<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Dave Linabury (Davezilla) on Medium]]></title>
        <description><![CDATA[Stories by Dave Linabury (Davezilla) on Medium]]></description>
        <link>https://medium.com/@davezillamedia?source=rss-57bfa919baf3------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*Wgcs9mZhjIHAAHBA175EgA.png</url>
            <title>Stories by Dave Linabury (Davezilla) on Medium</title>
            <link>https://medium.com/@davezillamedia?source=rss-57bfa919baf3------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 18:06:48 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@davezillamedia/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[I beg to differ, Neil]]></title>
            <link>https://medium.com/@davezillamedia/i-beg-to-differ-neil-166696cba524?source=rss-57bfa919baf3------2</link>
            <guid isPermaLink="false">https://medium.com/p/166696cba524</guid>
            <category><![CDATA[marketing-strategies]]></category>
            <category><![CDATA[oppenheimer-movie]]></category>
            <category><![CDATA[barbie-movies]]></category>
            <category><![CDATA[seo]]></category>
            <dc:creator><![CDATA[Dave Linabury (Davezilla)]]></dc:creator>
            <pubDate>Thu, 14 Sep 2023 18:59:36 GMT</pubDate>
            <atom:updated>2023-09-15T12:48:45.453Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JUdL5KjXhc1VYvy01s-GWQ.png" /><figcaption>OPPENHEIMER VS. BARBIE | IMAGE CREATED BY THE AUTHOR ON MIDJOURNEY 5.2</figcaption></figure><p><a href="https://neilpatel.com/">Neil Patel</a> is one of the most brilliant SEO professionals and marketers on the planet. I follow his work closely. Today, however, his newsletter left me wondering what he had been smoking. Here’s an excerpt:</p><blockquote>Both Oppenheimer and Barbie were major blockbusters. But Barbie blew Oppenheimer out of the water.</blockquote><blockquote>How?</blockquote><blockquote>The hot-pink diva had the perfect mix of marketing and timing on her side.</blockquote><blockquote>Both movies hit theaters on July 21. Students were on summer break. There was a nationwide heat wave and families were looking for air-conditioned activities.</blockquote><blockquote>And with a reported <a href="https://click.convertkit-mail.com/o8uv840mgqbqhkonxnvu2cpvkxwrr/z2hghnh387zv8gbp/aHR0cHM6Ly9hZHZhbmNlZC5ucGRpZ2l0YWwuY29tL2NvbnN1bHRpbmctcGFnZS13aXRoLXV0bXMvP3V0bV9tZWRpdW09ZW1haWwmdXRtX3NvdXJjZT1jb252ZXJ0a2l0LWxpc3QmdXRtX2NhbXBhaWduPXVzLW1rdC1jYW1wYWlnbnMtc3FsLXNlcnZpY2VzLWNvbnN1bHRpbmcmdXRtX2NvbnRlbnQ9Y29uc3VsdGluZyZ1dG1fdGVybT1zZXB0ZW1iZXItNi0yMDIzLWNvbnN1bHRpbmctZW1haWwtbmVpbHMtbGlzdA==">$150 million marketing budget</a>, Barbie had some brilliantly innovative campaigns supporting its release.</blockquote><p><strong>I disagree. </strong>Marketing is certainly critical. Hell, I work in it. I should know. But there are so many other factors at play here.</p><p>Full disclosure: I did not see <em>Barbie</em>. I did see <em>Oppenheimer</em> in a proper 70mm IMAX theater. I have, however, read enough about <em>Barbie</em> and people’s opinions on it and seen enough clips to make an educated guess as to why it did better.</p><h4>My take</h4><ul><li><em>Barbie</em> was upbeat, and the public wants upbeat and lighthearted things in the summer. <em>Oppenheimer</em> was sad, disturbing, and outright depressing at times.</li><li><em>Barbie</em> was about girl power; <em>Oppenheimer</em> was about military power.</li><li><em>Barbie</em> was funny; <em>Oppenheimer</em> was serious.</li><li><em>Barbie</em> didn’t require too much thinking (I’m assuming); <em>Oppenheimer</em> did.</li><li><em>Barbie</em> was shown in normal theaters; <em>Oppenheimer</em> was <strong>only</strong> shown in IMAX theaters. <strong>This alone is the biggest argument.</strong> Nolan chose to shoot for 70mm IMAX. That severely limited the number of potential screenings to just 19 theaters. Barbie, however, had the ability to be shown in 2,208 theaters. Even if you count the digital IMAX theaters, of which there are substantially more, it’s still less than half.</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=166696cba524" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[AI or not, artists will always be here]]></title>
            <link>https://medium.com/@davezillamedia/ai-or-not-artists-will-always-be-here-a37676a6bb83?source=rss-57bfa919baf3------2</link>
            <guid isPermaLink="false">https://medium.com/p/a37676a6bb83</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[generative-ai-tools]]></category>
            <category><![CDATA[generative-art]]></category>
            <category><![CDATA[meta]]></category>
            <category><![CDATA[talent]]></category>
            <dc:creator><![CDATA[Dave Linabury (Davezilla)]]></dc:creator>
            <pubDate>Tue, 18 Jul 2023 20:32:19 GMT</pubDate>
            <atom:updated>2023-07-18T20:32:19.581Z</atom:updated>
            <content:encoded><![CDATA[<h4>My old friend, <a href="https://www.zeldman.com/">Jeffrey Zeldman</a> posted a link to <a href="https://ai.meta.com/blog/generative-ai-text-images-cm3leon/">CM3leon</a>, Meta’s entry into the generative art ring.</h4><h4>And below is the image they chose to show off its capabilities:</h4><figure><img alt="Oh good god, no. This is so bad, you won’t want the description." src="https://cdn-images-1.medium.com/max/1024/1*D0u2W5bkT3LnpP16bVX7Ww.png" /></figure><p><strong>Utter gobshite. </strong>I am sure it can do better, but this? Really? A multi-billion dollar company thinks a sad cactus with pasted-on sunglasses is going to entice us to drop Midjourney for their model? The image looks like someone’s first failed attempt at Photoshop.</p><p>This post from Jeffrey led a former coworker (who is rather anti-AI) to write this:</p><blockquote>“Eventually this will lead to the commoditization of art to the point of boring.” <em>— Jim Amos</em></blockquote><p>I disagree. This post is a longer response to Jim’s quip.</p><p>The average Joe is <em>already</em> producing boring, offensive, blasé art, because he simply does not have <strong>good ideas</strong> to give to the LLM, so we’re seeing lots of flying Corgis, waifus popping out of their uniforms, and, sigh, childish cacti in poorly rendered deserts.</p><h3>Sturgeon’s Law:</h3><blockquote>“Ninety percent of science fiction is crud. But then ninety percent of everything is crud, and it’s the ten percent that isn’t crud that is important. And the ten percent of science fiction that isn’t crud is as good as or better than anything being written anywhere.”</blockquote><p>And indeed, a much smaller group of artists—who are not tech-averse — are producing AI art that is emotionally gripping, beautiful, haunting, and will stand the test of time.</p><figure><img alt="A ghostly entity sitting alone in a dusty room, hazy light streaming through." src="https://cdn-images-1.medium.com/max/1024/1*64YkqX3n_mTeLsz_un-1ug.png" /><figcaption>Image created in Midjourney by the author</figcaption></figure><h3>Creating great AI art requires several skills:</h3><h4>An idea</h4><p>This is unchangeable (at least for now). AI needs humans to generate ideas. I know you’re thinking, “But Dave, I’ve used ChatGPT to come up with ideas.” Yes, but they are all previously used ideas from other humans. Ideas that are well-known, overused, and not in any way original. Most are simply pulled from blog posts on heuristics and best practices. That’s fine and risk-averse, but unlikely to win you any innovation awards.</p><h4>Prompting skills</h4><p>These change according to the platform. Prompts that work on one do not always translate well to another. It reminds me of web design in the early 2000s, when we had to code differently for each browser. Eventually, <a href="https://www.webstandards.org/">coding standards emerged</a>. I predict we may see prompting standards emerge.</p><h4>Lighting, tonal, and form variations</h4><p>These are innate skills, that can be learned in art or photography school, and it takes years to master how to spot and correct them. Beginners in art rarely incorporate proper lighting subtleties like specular highlights or rim lighting. It’s still common to see even seasoned painters make rookie mistakes, such as adding starlight reflections to chrome instead of oval pops of light.</p><h4>Ability to spot and correct for hallucinations</h4><p>It’s not just looking for seven fingers or 18-pack abs. Hallucinations can be extremely subtle: cultural subtleties that might cause embarrassment or offense to someone. Do not assume Midjourney, Leonardo, et al, understand these cues. <strong>Just because an image looks realistic doesn’t mean it’s accurate.</strong></p><h4>Editing</h4><p>In Strunk &amp; White’s classic, <em>“The Elements of Style,”</em> there is this gem:</p><blockquote>“Vigorous writing is concise. A sentence should contain no unnecessary words, a paragraph no unnecessary sentences, for the same reason that a drawing should have no unnecessary lines and a machine no unnecessary parts. This requires not that the writer make all his sentences short, or that he avoid all detail and treat his subjects only in outline, but that every word tell.”</blockquote><h3>Final thoughts</h3><p>This is still the early days of generative AI. It will settle down eventually, and the real artists will emerge and produce works of great beauty that were previously unthinkable.</p><p>And average Joe will continue to use AI to make stupid memes about beer and butts, just like he has done with Photoshop for 30 years. Remember how bad the web looked in the late ‘90s? Same thing. Different code.</p><p>We have only begun to scratch the surface of what is possible. I welcome your thoughts.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a37676a6bb83" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Make jaw-dropping 360º Facebook panoramas from your Midjourney creations]]></title>
            <link>https://medium.com/@davezillamedia/make-jaw-dropping-360%C2%BA-facebook-panoramas-from-your-midjourney-creations-e9782a92c3fc?source=rss-57bfa919baf3------2</link>
            <guid isPermaLink="false">https://medium.com/p/e9782a92c3fc</guid>
            <category><![CDATA[panorama]]></category>
            <category><![CDATA[facebook]]></category>
            <category><![CDATA[midjourney-v5]]></category>
            <category><![CDATA[midjourney-how-to]]></category>
            <category><![CDATA[midjourney-prompts]]></category>
            <dc:creator><![CDATA[Dave Linabury (Davezilla)]]></dc:creator>
            <pubDate>Fri, 07 Jul 2023 15:56:06 GMT</pubDate>
            <atom:updated>2023-07-07T16:04:33.382Z</atom:updated>
            <content:encoded><![CDATA[<h4>Yes, it can be done. And it’s really, really easy.</h4><p>I am a sucker for the 360º panoramas on Facebook. I&#39;ve always wanted to make one from my art. With the welcome addition of <strong>Midjourney’s pan feature</strong> and one free app, my dream became a reality.</p><h4>Step 1. Make the initial image</h4><p>I’m sure you’ve used the <em>pan feature</em> by now if you&#39;ve read my posts, but in case you haven’t, here’s the quick tutorial.</p><p>Create an image. Here is the <strong>upscale</strong> of the image I chose and its prompt:</p><pre>a cinematic view of a dark medieval dungeon, The Focal, a group of magicians <br>in hooded black robes perform a ceremony, a mysterious woman in front of the <br>altar --ar 16:9 --style raw</pre><figure><img alt="A Midjourney image of a hooded woman at an altar in a medieval dungeon. She is kneeling, facing the viewer, and her hands are on her lap. She is surrounded by candles." src="https://cdn-images-1.medium.com/max/1024/1*QUbAQZotin4XPiSXdgBpmQ.png" /></figure><p>Notice that I used the aspect ratio of 16:9 for a widescreen image. This isn’t important, but if you want a wide pano, it will get you there quicker than a square will.</p><h4>Step 2. Add the first camera pan</h4><p>After upscaling, you’ll now see (on versions 5.2 and later) four arrows (left, right, up, and down). Choose a direction. I chose <em>pan right</em>. Note that once you choose a direction (left/right vs. up/down), you must continue in that direction. In other words, you can pan left and right, or up and down, but not left and up, for example.</p><figure><img alt="A screenshot from Midjournewy showing the new pan arrow buttons" src="https://cdn-images-1.medium.com/max/675/1*Mqy4ymNMIq2FlGvWJuYR4w.png" /></figure><p>And that gave me this image:</p><figure><img alt="Showing the newly panned image." src="https://cdn-images-1.medium.com/max/1024/1*yAXOHrJi-EWLa4urnVOM7w.png" /></figure><p>Notice how it added new characters? Now, as <a href="https://medium.com/generative-ai/how-to-tell-a-story-in-midjourney-with-pan-and-zoom-876f2e9c0050">Paul DelSignore pointed out in his post on panning</a>, you can use the remix feature to change the prompt to something new and create a story! His post is worth reading, (and he is someone you should be following) so definitely check it out. I chose not to change the prompt, and instead, to let Midjourney write its own story.</p><h4>Step 3. Keep panning until you are happy with it.</h4><p>Theoretically, you can keep panning until you run out of drive space to house the image. Here’s where I stopped (three right pans, and three left pans). Because I kept my aspect ratio at 16:9, the image expanded horizontally in no time.</p><figure><img alt="The final image with six pans." src="https://cdn-images-1.medium.com/max/1024/1*1Gbix_-Pr1CetmH33bFBNQ.png" /></figure><h4>Step 4. Make the pano</h4><p>This is the fun and easy part. You&#39;ve just done the hard part. And repeatedly tapping a button is not that hard, right? To make this into a panoramic image that you can view on Facebook or Google, you need a free app called <a href="https://exiffixer.com/online.htm#exif-fixer-online">Exif Fixer</a>. There is an online version you can upload to, or, if you think you will be making lots of these, download the free desktop app. It’s available for MacOS, Windows, and Linux.</p><p><strong>Exif is the invisible metadata your camera (or phone) adds to a digital image</strong> to identify what make of device was used, date, geolocation, resolution, etc. Panoramic images need to be taken with a “one-shot” camera, and Midjourney, as of this writing, cannot supply that type of metadata. <strong>That’s where this Exif Fixer comes in.</strong></p><p>Upload an image, keep the default settings until you know what you are doing (or read the short help popup).</p><blockquote><strong>Important note: </strong>You will need to convert your image to a JPG before using Exif Fixer. If you try it with a PNG, nothing will happen.</blockquote><figure><img alt="A screenshot of Exif Fixer’s online version." src="https://cdn-images-1.medium.com/max/661/1*AMgekhHDcLz9Xvzi6LkqwQ.png" /><figcaption>A screenshot of Exif Fixer’s online version.</figcaption></figure><p><strong>That’s it!</strong> Your image will be saved with new exif data that will fool Facebook, Google, and other pano-supporting websites into thinking it’s legit.</p><h4>Test it out.</h4><ol><li>Upload your new image to Facebook.</li><li>Before hitting POST, hit EDIT. You’ll see controls to pan around your image, like this, and the ability to choose your Starting View:</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/587/1*VLInlihZgbybHp8Cawo7Mg.png" /></figure><p>If it’s working, hit save and publish! Now all your friends can play with it.</p><h4>Hat tip to my friend <a href="https://www.linkedin.com/in/faith-kaminsky-cohen-1a5391b/">Faith Kaminsky </a>for showing me Exif Fixer.</h4><h3>Author’s note</h3><p><em>Thanks for reading. Don’t forget to leave a clap or 20, follow, share, and comment—you know the drill. If you liked this piece, I post on a strict, “when I get my sh*t together” basis. You can subscribe to be notified when I do.</em></p><h3>Not a Medium member? Join today!</h3><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e9782a92c3fc" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How to create ad-worthy automotive shots with AI]]></title>
            <link>https://medium.com/@davezillamedia/how-to-create-ad-worthy-automotive-shots-with-ai-8ea52a924490?source=rss-57bfa919baf3------2</link>
            <guid isPermaLink="false">https://medium.com/p/8ea52a924490</guid>
            <category><![CDATA[automotive]]></category>
            <category><![CDATA[midjourney-v5]]></category>
            <category><![CDATA[midjourney-prompts]]></category>
            <category><![CDATA[ai-art]]></category>
            <category><![CDATA[tutorial]]></category>
            <dc:creator><![CDATA[Dave Linabury (Davezilla)]]></dc:creator>
            <pubDate>Tue, 30 May 2023 21:35:07 GMT</pubDate>
            <atom:updated>2023-05-30T21:35:07.202Z</atom:updated>
            <content:encoded><![CDATA[<h4>Time was, creating realistic automotive shots with impressive lighting, perfect weather conditions, and flawless reflections was only in the realm of professional photographers.</h4><figure><img alt="A Lamborghini Avenator driving at 200 MPH on the Autobahn." src="https://cdn-images-1.medium.com/max/1024/1*WhITpSg9SmkqFfr7K1msGg.png" /><figcaption>LAMBORGHINI AVENATOR</figcaption></figure><h4>REQUIREMENTS:</h4><ul><li>A Discord and Midjourney account</li><li>A basic understanding of using Midjourney</li><li>Running at least version 5, preferably <strong>5.1 in RAW mode</strong> as it is less “opinionated” on how things should look</li></ul><p><strong>NOTE: </strong><em>I want to stress that while I use capitalization in the following lists, this is strictly for skimming. Midjourney doesn’t care about capitalization or grammar.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/794/1*BOLPQsHKp9xj1s_GN7Lj-w.png" /><figcaption>These are the settings I used for this article</figcaption></figure><h3>Automobiles + Midjourney</h3><h4>Automotive Prompt Anatomy</h4><p>I try to follow best practices and heuristics for prompting. I like to create prompts that are repeatable, predictable, and look professional. There are three elements that give the most consistent results:</p><ol><li>Subject</li><li>Styles</li><li>Parameters</li></ol><h3>Subject</h3><h4>Vehicle</h4><p>The <strong>subject</strong> includes, of course, the vehicle, and the more specific, the better. Include the trim style and year (or decade) whenever possible. The more exacting you are, the better the prompt.</p><p>For example, suppose you need a Ferrari Testarossa. Just using Ferrari as a prompt is OK, but who knows what you’ll get?</p><figure><img alt="Ferrari" src="https://cdn-images-1.medium.com/max/1024/1*Vfd-XZItghD4apbJlg9piQ.png" /><figcaption>/imagine prompt <strong>Ferrari</strong> — ar 16:9</figcaption></figure><p>Using Ferrari 512 TR is better, and spelling it out as Ferrari 512 Testarossa gives the best results:</p><figure><img alt="Ferrari 512 Testarossa" src="https://cdn-images-1.medium.com/max/1024/1*Z99NF0maGERENfYnTIgNBg.png" /><figcaption>/imagine <strong>Ferrari 512 Testarossa</strong> — ar 16:9</figcaption></figure><h4>Action</h4><p>The second part of the <strong>subject</strong> is its <strong>action</strong> (parked, driving, speeding, sliding, cruising):</p><figure><img alt="an Aston Martin Vanquish speeding on the Autobahn at 200 MPH" src="https://cdn-images-1.medium.com/max/1024/1*J5twUKxWqt4bsP9bQtTWig.png" /><figcaption>/imagine an Aston Martin Vanquish <strong>speeding </strong>on the Autobahn at 200 MPH — ar 16:9</figcaption></figure><p>Notice what a great job it did adding motion blur? I used to add the phrase <em>motion blur</em>, but giving it a specific speed does the same thing.</p><h4>Location</h4><p>The final part of the <strong>subject</strong> is its <strong>location</strong> (on a frozen lake, next to a volcano, on the Autobahn, deep in the Amazonian jungle).</p><figure><img alt="A dark gray Jeep Rubicon speeding through powder snow in Aspen, CO" src="https://cdn-images-1.medium.com/max/1024/1*s8ZkGpNfgQVE7ILDrYt-vg.png" /><figcaption>/imagine a dark gray Jeep Rubicon speeding t<strong>hrough powder snow in Aspen, CO </strong>— ar 16:9</figcaption></figure><p>Look at the realism of the snow and forest! The sunlight hitting the powder is even more impressive. We didn’t even ask it to include that detail, but that provides a nice segue into the next section: <strong>styles</strong>.</p><h3>Styles</h3><h4>Camera, Lenses, and Film Type</h4><p>Using a specific camera lens and film type will greatly increase the realism of the image.</p><h4>Camera Types</h4><p>If I missed any, please inform me.</p><ol><li><strong>Camcorder</strong> (grainy, has a distinct ’90s look)</li><li><strong>Disposable camera (</strong>35 mm lens, fixed focus)</li><li><strong>DSLR</strong> (I use this one frequently because of its exceptional resolution and clarity)</li><li><strong>GoPro</strong> (best for action shots and selfies as it has a fish-eye lens effect)</li><li><strong>Polaroid</strong> (contrasty but with a low black point, in square aspect ratio, it will even add a Polaroid border!)</li></ol><p>Let’s work with our Jeep Rubicon from the last section and add a simple camera type — the GoPro. Notice that the camera is put first. This is because the items mentioned first have <em>slightly</em> more prominence. This can be overridden with weighting, but that is a different topic that <a href="https://medium.com/ai-art-creators/the-importance-of-using-text-weights-in-midjourney-v5-f0d1e852278f">Paul DelSignore covered at length</a>.</p><figure><img alt="a GoPro shot of a dark gray Jeep Rubicon speeding through powder snow in Aspen, CO" src="https://cdn-images-1.medium.com/max/1024/1*3Hz-sWUM0ChlnK3IerfyTw.png" /><figcaption>/imagine <strong>a GoPro shot </strong>of a dark gray Jeep Rubicon speeding through powder snow in Aspen, CO, — ar 16:9</figcaption></figure><h4>Camera Models</h4><p>All types of specific camera models are available. I have seen too many prompts with the <a href="https://prompthero.com/search?q=Canon+EOS+Rebel">Canon EOS Rebel</a> or various <a href="https://prompthero.com/search?model=Midjourney&amp;q=Nikon&amp;source=473e8455f2b">Nikon</a>s. This is likely due to beginners copying prompts from more experienced creators. Try some quirky ones, like a Lomo, a Lytro, the Sakura Petal, a Rollei 35, or a Stylophot. Try expensive cameras like a Hasselblad H6D-400C MS or a Leica 0-series no. 122. Go wild.</p><h4>Film Types (includes movie film types)</h4><p>There’s likely more than this list. Please add them in the comments.</p><ol><li>Afga Vista 200</li><li>CinemaScope</li><li>Daguerreotype plates</li><li>Fomapan 400 (great retro look)</li><li>Fujifilm: Pro and Superia</li><li>Ilford XP2 400</li><li>Instax</li><li>Kodak: 400TX, Ektar, Gold, Infrared, and Portra</li><li>Lomo</li><li>Technirama</li><li>TechniColor</li><li>Tintype (also melainotype or ferrotype)</li><li>Tri-X 400</li><li>Velvia</li><li>VistaVision</li></ol><h4>Camera Shots and Angles</h4><p>AI Tuts has <a href="https://aituts.com/midjourney-camera-prompts/#Camera_Angles">an in-depth article about camera angles</a> that I highly recommend. Here is their list, minus the descriptions.</p><ol><li>360 view</li><li>Back-angle</li><li>Bird-eye view (also called drone-angle or drone-view)</li><li>Cinematic (very popular)</li><li>Closeup shot</li><li>Eye-level shot</li><li>Far-shot angle</li><li>Full-body shot</li><li>Full-shot angle</li><li>Glamour shot</li><li>Ground-shot angle</li><li>High-angle</li><li>Low-angle</li><li>Macro-shot</li><li>Medium-full shot</li><li>Medium-shot angle</li><li>Satellite-angle</li><li>Side-angle (or side-view, lateral view)</li><li>Waist-up shot</li><li>Wide-angle shot</li><li>Worms-eye view</li><li>Ultra-wide angle</li></ol><h4>Lighting</h4><p>Lighting alone can be an entire series of articles. Here is <a href="https://medium.com/geekculture/a-guide-to-lighting-in-midjourney-3c94d00b20d0">a great resource on Midjourney lighting</a> from Paul DelSignore. One of the most important takeaways from Paul’s article is this:</p><blockquote><strong>Don’t use photography terminology.</strong> When I started rendering portraits, I tried using terms in my prompts like <em>split lighting</em>, <em>Flatlight</em>, and <em>Butterfly lighting</em> but they were not recognized in Midjourney. Instead, use <strong><em>natural language descriptions</em></strong> of the lighting effect you want.</blockquote><blockquote>For example, instead of using a term like ‘<em>Backlight</em>,’ use a phrase like <em>‘light from behind</em>’ or ‘<em>lighted background</em>.’ If you use ‘<em>Butterfly lighting</em>” you will get butterflies in your image :)</blockquote><h4>The Big List of Lighting Options</h4><p>A resource you <strong>must</strong> bookmark is <a href="https://github.com/willwulfken/MidJourney-Styles-and-Keywords-Reference/blob/main/Pages/MJ_V5/Style_Pages/Just_The_Style/Lighting.md">this list on GitHub</a>, as it not only lists all the possible lighting options but also has visual examples of each. It has sections on:</p><ol><li>Types of lights</li><li>Lamps and tubes</li><li>Types of displays</li><li>Lighting styles and techniques</li><li>Global illumination</li><li>Shadows</li></ol><p>Suffice it to say, you should experiment with combinations of the following lighting options to get a really good shot:</p><ol><li><strong>Time of day: </strong>dawn, dusk, noon, high noon, daylight, nighttime, golden hour, midday, blue hour, sunrise, sunset, midnight, etc.</li><li><strong>Weather conditions: </strong>cloudy, earthquake, foggy, hazy, hurricane, rain, snow, storm, sunny, tempest, etc. Note that you can use any form of these words (ex., rain, raining, rainy) and get similar results.</li><li><strong>Lighting direction:</strong> chrome lighting, left side lighting, overhead lighting, right side lighting, underneath lighting, volumetric lighting, etc.</li></ol><h3>Putting them all together</h3><p>Let’s try some images in a few different combinations to give you an idea of what’s possible.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*wXB5EMaLIhKpR3NdUT9J6g.png" /><figcaption>/imagine <strong>a cinematic shot of a Subaru Outback driving through sunny, rural America, shot with a Lomo, golden hour</strong>, — ar 16:9</figcaption></figure><p>Let’s try a different vehicle, weather conditions, and camera gear.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*fYdU-cXtbYVPpogz-xWteg.png" /><figcaption>/imagine <strong>an overhead shot of a 1966 Karmann Ghia Spider driving through a rainstorm, heavy rains, dim lighting, misty gothic</strong> — ar 16:9</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*35CbnIhsNE3lmym_sXdktg.png" /><figcaption>/imagine <strong>a 1963 Corvette Stingray speeding through a future, cyberpunk Detroit, BladeRunner 2049 styling, neon lights, nighttime</strong>, — ar 16:9</figcaption></figure><p>Let’s ignore the fact that GM Legal would have a heart attack seeing one of their vehicles without a driver. But look at the lighting effects in this. Cyberpunk is a great term to use to add hot, bright lights to an image as well as a slightly dystopian future scene. Adding “BladeRunner 2049” to the prompt gave it rain and a Neo-Tokyo feel without going full-Manga. All of the lights are perfectly reflected all over the vehicle body and rims in a manner that would take dozens of hours to achieve using Photoshop.</p><p>One more…</p><figure><img alt="a BMW 2002 driving at top speed while the road explodes from underneath" src="https://cdn-images-1.medium.com/max/1024/1*ap6zzEg_36bpsibLaNOZbg.png" /><figcaption>/imagine <strong>film still, cinematic street scene, a BMW 2002 driving at top speed while the road explodes from underneath, volumetric lighting, </strong>— ar 16:9</figcaption></figure><p>Yet another driverless car. Sigh…</p><p>We are not going to cover <strong>parameters</strong> in this article, as that is a <em>huge</em> discussion. Instead, <a href="https://stefvanlooveren.me/blog/listing-all-midjourney-parameters">I direct you here</a>, dear reader.</p><h3>Now a warning…</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_c8gJ83DemqAFElBdYEnSw.png" /></figure><h4>#NotAllCars</h4><p>I showed you some great examples, but despite everything you’ve learned here, do not promise anyone they can get any car they want. <strong>It isn’t possible. </strong>This is a limitation of what Midjourney has been trained on. For example, just try to get Midjourney to create an older Saab that doesn’t end up looking like PeeWee Herman designed it.</p><p>Before promising, just say, “Let me look into that,” then try it out on Midjourney. Give yourself 4–5 vehicle models to try before calling it quits. No one wants that conversation if Midjourney can’t get it right.</p><h4>Not My Type</h4><p>Secondly, type. It’s an issue. As of this writing, you will not get type or logos to render properly. Type isn’t understood by AI art apps, except to a certain extent, by Adobe Firefly, an app that can create incredible headlines and logo marks made of snakeskin, bricks, or whatever you can think of. However, even Firefly cannot render type in its images.</p><h3>Author’s note</h3><p><em>Thanks for reading. Don’t forget to clap, follow, share, and comment — you know the drill. If you liked this piece, I post on a strict, “when I get my sh*t together” basis. You can subscribe to be notified when I do.</em></p><h3>Not a Medium member? Join today!</h3><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8ea52a924490" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Power your prompts with some mind-blowing Midjourney magic]]></title>
            <link>https://medium.com/@davezillamedia/want-to-try-something-wild-on-midjourney-ac163403fc32?source=rss-57bfa919baf3------2</link>
            <guid isPermaLink="false">https://medium.com/p/ac163403fc32</guid>
            <category><![CDATA[forbiddenprompts]]></category>
            <category><![CDATA[emoji]]></category>
            <category><![CDATA[midjourney-prompts]]></category>
            <category><![CDATA[ai-art-prompts]]></category>
            <category><![CDATA[creativity]]></category>
            <dc:creator><![CDATA[Dave Linabury (Davezilla)]]></dc:creator>
            <pubDate>Sun, 28 May 2023 22:03:29 GMT</pubDate>
            <atom:updated>2023-05-29T17:42:25.022Z</atom:updated>
            <content:encoded><![CDATA[<p>While I’ve never been bored using Midjourney, it’s easy to fall into a rut of prompting a certain way every time. Sometimes you just gotta shake things up. I have some quirky ideas that are guaranteed to get you some unexpected and really cool results.</p><h4>Emoji prompts</h4><p>Emojis are fun to add to a prompt, but have you ever made the emoji the sole prompt? As in: /imagine ⚛️ — s 1000 — c 50. Here’s what that prompt gave me:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qxLNoLQSTJOQ45uJ9hrJaw.png" /><figcaption>/prompt ⚛️ — s 1000 — c 50</figcaption></figure><p>Then I tried a few more…</p><h4>Head exploding emoji 🤯</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*X5r4lzwP8K28v-Lp86XGFQ.png" /><figcaption>0/imagine 🤯 — s 1000 — c 50</figcaption></figure><h4>Dotted line face emoji 🫥</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*pkO1ZjCSEGPBmzxeH1MyvQ.png" /><figcaption>/imagine 🫥 — s 1000 — c 50</figcaption></figure><p>I have the feeling Midjourney sees the emojis and just makes shit up. I mean, look at the shrug emoji:</p><h4>Shrug emoji 🤷‍♂️</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Z5NUuQTDaFVmJhfyUdxG5g.png" /><figcaption>/imagine 🤷‍♂️ — s 1000 — c 50</figcaption></figure><h4>The fault in our emojis</h4><p>Then I had another idea. What would Midjourney make of the astrological emojis? Turns out, unless you add “man” at the end, you will only get women. And unless you specify a race, you’re getting a Caucasian woman. Here are the first three sun signs with only the astrological emoji as a prompt.</p><h4>Aries</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*pDby8elQ5z32KMQRN7m9fw.png" /><figcaption>/prompt ♈️ — s 1000 — c 50</figcaption></figure><h4>Taurus</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*f1_CodmPu-nhzWoZEaXtdg.png" /><figcaption>/prompt ♉️ — s 1000 — c 50</figcaption></figure><h4>Gemini</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/730/1*pj4B249qdiHxEBkWypBulQ.png" /><figcaption>/prompt ♊️ — s 1000 — c 50</figcaption></figure><p>I made all 12 signs, but you get the idea.</p><h4>Shameless prompting</h4><p>I have used the moniker Davezilla since 1994. Don’t try to steal it; there is a <a href="https://en.wikipedia.org/wiki/Davezilla.">Wikipedia page about me </a>and <a href="https://www.urbandictionary.com/define.php?term=Davezilla%27d">I’m a verb in the Urban Dictionary</a> which makes me officially Internet-Old. The reason I mention that I am established is because there are enough mentions of me online (literally tens of thousands) that Midjourney kind of knows who I am.</p><p>Back to the prompt. Facebook covers need to be 831 pixels X 315 pixels, which I rounded to an aspect ratio of 8:3. It fits perfectly, by the way. So the prompt I tried was:</p><pre>/imagine make a funny collage that Davezilla would be proud of. --ar 8:3 --s 1000 --c 50</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*fdcY7b6KJzTo-q7rfDtnHA.png" /></figure><p>I am proud. So proud. It is now my Facebook cover. Now imagine swapping my name with someone you admire. Try Salvador Dali, Henry Rollins, or Neil Gaiman. It’s a far different result than if you ask for “in the style of.”</p><p>Let me know what you come up with!</p><h3>Author’s note</h3><p><em>Thanks for reading. Don’t forget to clap, follow, share, and comment — you know the drill. If you liked this post, I post on a strict, “when I get my sh*t together” basis. You can subscribe to be notified when I do.</em></p><h4>Not a Medium member? Join today!</h4><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ac163403fc32" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[I will be on CBS News tonight (again)]]></title>
            <link>https://medium.com/@davezillamedia/i-will-be-on-cbs-news-tonight-again-8ec76de8ca6d?source=rss-57bfa919baf3------2</link>
            <guid isPermaLink="false">https://medium.com/p/8ec76de8ca6d</guid>
            <category><![CDATA[cbs-news]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[copyright-law]]></category>
            <category><![CDATA[law]]></category>
            <dc:creator><![CDATA[Dave Linabury (Davezilla)]]></dc:creator>
            <pubDate>Mon, 22 May 2023 17:34:24 GMT</pubDate>
            <atom:updated>2023-05-22T17:34:24.158Z</atom:updated>
            <content:encoded><![CDATA[<p>Tonight we will be discussing the pros and cons of AI copyright law. We’ll be on in the 7 p.m. EDT slot. Hope to see you!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*iR0ZohFDPltCWhkcJEyG_Q.jpeg" /><figcaption>That’s me in the corner. That’s me in the spotlight, losing my hairline.</figcaption></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8ec76de8ca6d" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[I put my dog’s DNA into Midjourney. OMG. It nailed it.]]></title>
            <link>https://medium.com/@davezillamedia/i-put-my-dogs-dna-into-midjourney-omg-it-nailed-it-b60b8c92fc8d?source=rss-57bfa919baf3------2</link>
            <guid isPermaLink="false">https://medium.com/p/b60b8c92fc8d</guid>
            <category><![CDATA[midjourney-prompts]]></category>
            <category><![CDATA[pets]]></category>
            <category><![CDATA[experimentation]]></category>
            <category><![CDATA[midjourney-v5]]></category>
            <category><![CDATA[dna-test]]></category>
            <dc:creator><![CDATA[Dave Linabury (Davezilla)]]></dc:creator>
            <pubDate>Fri, 19 May 2023 20:10:28 GMT</pubDate>
            <atom:updated>2023-05-19T20:36:50.442Z</atom:updated>
            <content:encoded><![CDATA[<h4>We tried something crazy. What if we fed our dog’s DNA percentages in as prompt weights? Would Midjourney be able to recreate our dog?</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ot1dIObALTcae3ulLHtaZw.jpeg" /><figcaption>EMMA!</figcaption></figure><h4>My wife and I have an adorable mutt named Emma.</h4><p>She’s a rescue, and we always wondered what her makeup was. We were told she was part Dachshund and part Chihuahua, but we thought there was more to it. So we had her DNA tested, and sure enough, she is a mix of <em>at least </em>seven dogs.</p><p><em>“What does this have to do with Midjourney, Dave?” </em>I&#39;m glad you asked.</p><p>After reading <a href="https://medium.com/@pdelsignore">Paul Signore’s</a> post, <a href="https://medium.com/ai-art-creators/the-importance-of-using-text-weights-in-midjourney-v5-f0d1e852278f">The Importance Of Using Text Weights In Midjourney V5</a>, I tried some experiments. One was particularly successful, so I mentioned it to Paul in his comments, and he said it was a good use case.</p><p>The experiment was to feed all seven parts of Emma’s genetic makeup into Midjourney, use weighting to equal the percentages of each breed, and see how close it gets to what she actually looks like.</p><h4>Here is Emma’s genetic makeup:</h4><ul><li>Chihuahua, 29%</li><li>Dachshund, 16%</li><li>Toy Fox Terrier, 15%</li><li>Russell Terrier, 12%</li><li>Rat Terrier, 9%</li><li>Shitzu, 8%</li><li>Basset Hound, 5%</li><li>Mutt, 6%</li></ul><p>Each percentage became a weight, so <em>29% Chihuahua </em>became Chihuahua::29, and so on.</p><h4>And here is the prompt I wrote:</h4><pre>A realistic depiction of a dog that is a mix of Chihuahua::29 <br>Dachshund::16 Toy Fox Terrier::15 Russell Terrier::12 Rat Terrier::9 <br>Shitzu::8 Basset Hound::5 Mutt::6 --s 1000 --c 50 --style raw</pre><p>How did it do? Pretty amazing, actually. Except for her coat color, it looks a <em>lot</em> like Emma!</p><h4>First, the real Emma:</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*4uCXI9DIh85jiJJLAZch1g.jpeg" /><figcaption>The real Emma. Such a cute pup.</figcaption></figure><h4>How Midjourney handled the DNA inputs:</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*mbe7IjXfVvsb8B9mTYFCEw.jpeg" /><figcaption>The lower right image looks just like her.</figcaption></figure><p>After some tweaking, we got the coat <em>closer</em>, but never got the all-over fawn color coat she has.</p><figure><img alt="Close! The ears and long body are spot on." src="https://cdn-images-1.medium.com/max/1024/1*DUNMT63T4G2h1wOojrZSGQ.png" /><figcaption>Close! The ears and long body are spot on.</figcaption></figure><p>Except for the coat, that is pretty spot on.</p><p>Have you tried this with your pets? Or with your own DNA (I don’t recommend this, but maybe making up DNA percentages might be interesting)? I am curious to hear your thoughts!</p><h4>Author’s note</h4><p><em>Thanks for reading. Don’t forget to clap, follow, share, and comment — you know the drill. If you liked this post, I try to post a few times a week. You can subscribe to be notified. Cool kids leave tips.</em></p><h4>Not a Medium member? Join today!</h4><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b60b8c92fc8d" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How to use the Describe prompt in Midjourney]]></title>
            <link>https://medium.com/@davezillamedia/how-to-use-the-describe-prompt-in-midjourney-95bf2ed32d20?source=rss-57bfa919baf3------2</link>
            <guid isPermaLink="false">https://medium.com/p/95bf2ed32d20</guid>
            <category><![CDATA[describe]]></category>
            <category><![CDATA[midjourney-v5]]></category>
            <category><![CDATA[midjourney-describe]]></category>
            <category><![CDATA[midjourney]]></category>
            <category><![CDATA[midjourney-prompts]]></category>
            <dc:creator><![CDATA[Dave Linabury (Davezilla)]]></dc:creator>
            <pubDate>Fri, 14 Apr 2023 15:23:02 GMT</pubDate>
            <atom:updated>2023-05-19T19:24:39.428Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/872/1*LcbCqPB2R-N8PVUyeTCl1A.png" /></figure><h3>Midjourney’s new <strong>Describe</strong> prompt will speed up your workflow, teach you better prompting techniques, and require fewer fast hours.</h3><h4>Here’s how:</h4><p>Describe was a godsend when it came out. No more jumping to ChatGPT to get the right prompt. And let’s face it, those novel-length prompts from OpenAI were often no better than a simple one. Of course, that depends entirely on your use case, and there are some instances when ChatGPT is absolutely the way to go. For most of us, <em>Describe is our new best friend</em>.</p><h3>What does it do?</h3><p>/Describe takes a photo and uses it as a base reference. I know what you’re thinking, “I could do that already with /imagine, Dave!” True, but Midjourney didn’t write the prompt for you, did it? That’s what /describe does. It writes four prompts for you, based on the image provided. You can use them as is, or add your own mods.</p><h3>How do I use /describe?</h3><h4>Step One</h4><p><strong>Choose an image. </strong>I found this cool fairycore pic on Pinterest.</p><figure><img alt="A beautiful, tattooed woman with blue hair, in a forest. Fairycore vibe." src="https://cdn-images-1.medium.com/max/1024/1*u31zI-Bpl9lAwuUh8-4nKw.jpeg" /><figcaption>From Pinterest.com</figcaption></figure><h4>Step Two</h4><p>Instead of using /imagine, you <strong>type /describe</strong>, like this:</p><figure><img alt="Typing /describe into Midjourney will prompt you for an image." src="https://cdn-images-1.medium.com/max/1024/1*5ib3IQyVj6hWjTPGHGeOQQ.png" /><figcaption>Typing /describe into Midjourney will prompt you for an image.</figcaption></figure><p>Notice how it immediately prompts for an image to be added. Just drag and drop one. It’s that simple. You’ll notice that the image URL is automatically added. No more right-clicking to copy the image address!</p><figure><img alt="Notice how the dropped image’s URL has already been added to the prompt. No cut and paste required." src="https://cdn-images-1.medium.com/max/1024/1*tpxioCFPOUAvspBSZIMnBQ.png" /><figcaption>Notice how the dropped image’s URL has already been added to the prompt. No cut and paste required.</figcaption></figure><h4>Step Three</h4><p>Hit Enter and the magic happens. Four prompts describing the image, the surroundings, lighting, style, and colors appear.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/674/1*toVLbDGUE3mgkNUBc12FAw.png" /></figure><h4>Step Four (final step)</h4><p>Pick a prompt and tap the corresponding button. Let’s see what the first button [1] does:</p><h4>Prompt [1]</h4><blockquote>a beautiful tattooed lady is sitting in the forest with bright blue hair and blue tattoos, in the style of floral accents, light teal and light green, earthcore, elaborate costumes, sony fe 12–24mm f/2.8 gm, photographically detailed portraitures, vibrant florals — ar 4:5 — v 5 —</blockquote><figure><img alt="a beautiful tattooed lady is sitting in the forest with bright blue hair and blue tattoos" src="https://cdn-images-1.medium.com/max/1024/1*BALFkcP5rxSZQLjRmXDBRA.png" /></figure><p>Not bad. Here’s what the other three gave me:</p><h4>Prompt [2]</h4><blockquote>blue tattooed women at the forests, in the style of fairytale-inspired, vibrant florals, environmental portraiture, foampunk, mallgoth, close up, pictorial — ar 4:5</blockquote><figure><img alt="blue tattooed women at the forests, in the style of fairytale-inspired, vibrant florals" src="https://cdn-images-1.medium.com/max/1024/1*tHMfeA8cz558P2ozKokG3A.png" /><figcaption>Two for the price of one.</figcaption></figure><p>Interesting. It wrote <em>women</em> and gave us consistently two women. These are really cool.</p><h4>Prompt [3]</h4><blockquote>tattoo woman with blue hair, costume, flower headgear, in the style of nature-inspired camouflage, kelly sue deconnick, evocative environmental portraits, emerald, 32k uhd, jessie arms botke, master of ink — ar 4:5</blockquote><figure><img alt="tattooed woman with blue hair, costume, flower headgear, in the style of nature-inspired camouflage" src="https://cdn-images-1.medium.com/max/1024/1*MEgdjecgN6Ze-NOus0fbEQ.png" /><figcaption>This time, the forest is taking over more, and she looks like a Flapper Elf.</figcaption></figure><h4>Prompt [4]</h4><blockquote>full body tattoo designs by emily dillard, in the style of evocative environmental portraits, green and azure, tokina opera 50mm f/1.4 ff, fairytale-inspired, margaret olley, 32k uhd, sublime wilderness — ar 4:5</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*iEFq95w3nXzk-B1sXLla1w.png" /><figcaption>Why is she a redhead now?</figcaption></figure><p>This one is the least successful, as her hair is red now, but maybe the Margaret Olley styling did that? In any event, none of these images have crazy artifacts or additional anatomical features.</p><p>I hope this was helpful for you. BTW, here’s the image I chose:</p><figure><img alt="My final image is a beautiful tattooed lady is sitting in the forest with bright blue hair and blue tattoos, in the style of floral accents, light teal and light green, earthcore, elaborate costume" src="https://cdn-images-1.medium.com/max/960/1*XtYAdZfOAoQzMvoU7eXjtg.png" /><figcaption>Eye-tracking is a bit off, but that’s an easy Photoshop tweak.</figcaption></figure><p>One thing may have crossed your mind: some of these prompt ideas are pretty good! Yes, and what’s more, <strong>the prompts from /describe give you an insight into how Midjourney interprets text</strong>, and what kind of text it likes. That should influence how you post.</p><h4>These were all done in Midjourney v5.1.</h4><p>All the articles I’ve read state clearly that MJ v5.1 prefers sentences to comma-separated text, and yet comma-separated is what MJ gave me. YMMV.</p><p>I hope this helps, and I’d love to hear what prompts it gave you. If you liked this post, consider subscribing to me so you know when each post comes out.</p><h3>Author’s note</h3><p><em>Thanks for reading. Don’t forget to clap, follow, share, and comment — you know the drill. If you liked this post, I post 2-3 times weekly. You can subscribe to be notified. Cool kids leave tips.</em></p><h4><strong>Not a Medium member? Join today!</strong></h4><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=95bf2ed32d20" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Bing Image Creator gets better?]]></title>
            <link>https://medium.com/@davezillamedia/bing-image-creator-gets-better-60b87ba4d9ca?source=rss-57bfa919baf3------2</link>
            <guid isPermaLink="false">https://medium.com/p/60b87ba4d9ca</guid>
            <category><![CDATA[ai-art]]></category>
            <category><![CDATA[midjourney]]></category>
            <category><![CDATA[bing-image-creator]]></category>
            <category><![CDATA[prompt]]></category>
            <category><![CDATA[promptcraft]]></category>
            <dc:creator><![CDATA[Dave Linabury (Davezilla)]]></dc:creator>
            <pubDate>Wed, 29 Mar 2023 13:58:29 GMT</pubDate>
            <atom:updated>2023-04-16T16:06:14.309Z</atom:updated>
            <content:encoded><![CDATA[<p>I wanted to write a followup piece to <a href="https://medium.com/@davezilla.art/bing-image-creator-first-impressions-f0e30d241d99">my article reviewing Bing Image Creator</a> (BIC). If you haven’t read it, you might want to first, but it’s not required. Here’s the TL;DR: I tried BIC, with no preconceptions. My first result was realistic, albeit disappointing in that it was a tad biased toward blonde, white women. I tried giving it a prompt to create a Nigerian goddess, the Orisha, Yemoja. Midjourney made some spectacular images of Yemoja, like this one:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/1*aeE_uUmVUTmoPzScdThWeg.png" /></figure><p>Here’s what BIC gave me with a similar, but shorter prompt than what I gave Midjourney.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*C2uu1c6-hH3j25qmGFbM5w.jpeg" /></figure><p>Not so good. Why did I use a shorter prompt? Because when I gave BIC the full, paragraph-long prompt, my prompt was flagged immediately and BIC refused to process the image. Sus, very sus.</p><p>I tried again today. This time, I asked for a very specific thing: A helicopter. And not just any helicopter. Here’s the prompt:</p><pre>an unmarked black UH-60 Black Hawk helicopter is circling overhead, the <br>helicopter is seen from a worms eye view, the sky is sunny and has a few <br>cumulus clouds, Canon EOS 5D Mark IV DSLR, f/5.6 aperture, 1/125 second <br>shutter speed, ISO 100, creative, expressive, unique, high-quality</pre><p>And here’s what I got:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/758/1*lZzDKLmBzTwVsYZoYnQWAw.png" /><figcaption>UH-60 Black Hawk helicopters, as requested.</figcaption></figure><p><strong>Not bad at all! </strong>There’s some wonkiness on the landing gear, which appear to be made from stretchy rubber, and the top left copter is losing parts midair, but overall, I am much more impressed. It’s also rare to see a worm’s eye view in AI art done well.</p><p>So I am changing my mind about BIC, but just a little. I am still unimpressed with the Orisha results, but I did manage to get it to create a decent black woman (still over-sharpened) so the fears of a biased-toward-white-people database I had in my last post are gone.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*6XDQE5N97gfl1j9cyPwSwQ.jpeg" /></figure><p>Have any of you tried Bing Image Creator? What was your experience? Let me know.</p><h3>Author’s note</h3><p><em>Thanks for reading. Don’t forget to clap, follow, share, comment — you know the drill. If you liked this post, I post 2–3X weekly. You can subscribe to be notified. Cool kids leave tips.</em></p><p><strong>Not a Medium member? Join today!</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=60b87ba4d9ca" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Bing Image Creator. First impressions.]]></title>
            <link>https://medium.com/@davezillamedia/bing-image-creator-first-impressions-f0e30d241d99?source=rss-57bfa919baf3------2</link>
            <guid isPermaLink="false">https://medium.com/p/f0e30d241d99</guid>
            <category><![CDATA[first-impressions]]></category>
            <category><![CDATA[bing-image-creator]]></category>
            <category><![CDATA[ai-art-generator]]></category>
            <category><![CDATA[ai-art]]></category>
            <category><![CDATA[midjourney]]></category>
            <dc:creator><![CDATA[Dave Linabury (Davezilla)]]></dc:creator>
            <pubDate>Wed, 22 Mar 2023 16:59:39 GMT</pubDate>
            <atom:updated>2023-03-22T16:59:39.397Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="A very basic, white chick, who just happens to have pronounced canines." src="https://cdn-images-1.medium.com/max/1024/1*PkEL7HJwiqza2lr2g_9oGg.jpeg" /><figcaption>Maybe it’s vampirism, maybe it’s Maybelline.</figcaption></figure><p><strong>After my friend </strong><a href="https://www.linkedin.com/post/edit/7044326446038732801/?author=urn%3Ali%3Afs_miniProfile%3AACoAAAAbXc8BHNAZvjA9-rDvSh_ffWBWIIZLvgI#"><strong>Chris Brogan</strong></a><strong> posted a cool AI image of himself as a crayon drawing made with </strong><a href="https://www.bing.com/create"><strong>Bing Image Creator,</strong></a><strong> I decided to try Bing’s newest toy out. After all, I’m a big fan of the direction AI Art is going. AI can’t replace me as an illustrator, but I see it as a helpful tool.</strong></p><p><strong>Math teachers in the 1970s were upset by the advent of calculators.</strong> “How will kids learn math?” Decades later, all kids still use calculators (apps) and learn their maths just fine. The same will happen with ChatGPT and Midjourney et al. English teachers are suffering a self-imposed, collective wedgie right now. If they’d simply take a deep breath, try the tools themselves — and really try to push them — they will realize AI is not a threat; it’s just another productivity tool.</p><p><strong>Back to Bing.</strong></p><p>I tried the recommended first prompt. They said “Try the prompt ‘surprise me’,” which I did. It gave me four images of women looking surprised (quelle shocker).</p><h3>My first impressions of Bing Image Creator</h3><p>The images looked impressive for the first three seconds, then I noticed what I hope is not the “Bing look” (you know how you can always tell a Midjourney image? Yeah, that). Let’s start with this one:</p><figure><img alt="This is an image of a woman looking surprised generated by Bing Image Creator. It’s fairly realistic if you ignore the fact that she appears to be eating fingers that are not her own." src="https://cdn-images-1.medium.com/max/1024/1*107OOu93Ys4oAQiuczXBGg.png" /><figcaption>Tap image for full size.</figcaption></figure><p><strong>OVER-SHARPENED: </strong>The images look like a beginner using Photoshop tried the Sharpen tool for the first time and went to town. Over-sharpened is not good.</p><p><strong>ANATOMICAL ISSUES: </strong>I expected these. After all, Midjourney still has them. I saw claims that Midjourney 5 had taken care of the hands issue, but I’ve seen some MJ5 images with extra fingers and toes, and Bing is not exempt from these artefacts either.</p><p><strong>A/S/L:</strong> I did not specify age, race, or a gender, yet my first experience with the tool was too obvious. I got four images of blonde, white women, aged 18–24.</p><figure><img alt="Four very blonde, very white, basic girls created by Bing Image Creator." src="https://cdn-images-1.medium.com/max/1024/1*hPVsWkIeSDb1cCUd0e3sqA.jpeg" /><figcaption>“But my fingers go to 11.”</figcaption></figure><p>As someone who has long practiced accessibility and inclusive design, not to mention, I work for an organization that helps black SMBs with their businesses, I am acutely aware of how many times white people are portrayed than others.</p><p>For a first impression. I wasn’t sure what to expect, honestly, but I will give Midjourney props in that if you don’t specify age, gender, or race, you will often get a nicely mixed bag.</p><p><strong>PROMPTS: </strong>I have only tried two images so far and the second one was immediately put up for review without creation. It was not anything nefarious or NSFW. I asked for a drawing of a goddess and got a message that it was flagged for review. My guess is, I am used to long, detailed prompts and Bing’s creator was only looking for a prompt of the simple, “Frog with an umbrella eating waffles” variety. Here is my exact prompt, which I had successfully used in Midjourney a week prior.</p><blockquote>The Nigerian Orisha Yemaya is standing on top of the sea. The waves are turbulent and the sky is dark and stormy. Yemaya is a tall and regal Yoruba Queen goddess with a red coral crown and plaited hair, She wears a dark blue dashiki. She has piercing black eyes and owns all of the fish in the sea. She is the owner of the ocean and of all water. She is beautiful and terrifying to behold. DSLR, natural lighting, photorealism, octane render.</blockquote><p>Here is how Midjourney interpreted that prompt:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/1*aeE_uUmVUTmoPzScdThWeg.png" /><figcaption>Dead on representation of the Orisha.</figcaption></figure><p>Midjourney nailed that prompt. Bing Image Creator put it up for review. Hmm. Again, it’s super early, so hard to tell why. I am hoping one of you may have insight into why that prompt was flagged.</p><p>I tried a second prompt and had success. This time, I tried a brief, natural language prompt:</p><blockquote>Bigfoot running away from a photographer in downtown Portland.</blockquote><figure><img alt="Four images of Bigfoot (or is it Bigfeet?) generated by Bing Image Creator. One of the Sasquatches has three legs!" src="https://cdn-images-1.medium.com/max/752/1*fQTlOFPEaC4m6AdcuaRaCw.png" /><figcaption>The photographer is more elusive than Bigfoot.</figcaption></figure><p>Not bad, if you ignore the obvious anatomical issues (like the upper right image having three legs, and a few of the Sasquatches having extra digits). To be fair, it’s better than all of the Bigfoot images I have tried on other AI text-to-image generators.</p><p>I use Bigfoot intentionally as a test object because:</p><ol><li>Bigfoot has never been caught (if it exists) so a lot of guesswork is placed on the system.</li><li>Bigfoot is covered in fur, according to witnesses, which can be problematic for systems to render.</li><li>Bigfoot has an anatomy and posture similar to humans, making it an interesting challenge.</li></ol><p>I browsed through the BIC gallery of example images. Many were designed to look as if they were “painted in a classical art style.” They do not. They look like Adobe’s first attempt at making fine art filters back in the late 1980s. Ouch.</p><h3>Summary</h3><p>So I need to try several more prompts that (crosses fingers) won’t be flagged before I make a definitive statement, but at this point, my impression is that if BIC can only use short prompts, it’s a high-end toy — not a high-end tool.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f0e30d241d99" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>