<?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 Rob Peart on Medium]]></title>
        <description><![CDATA[Stories by Rob Peart on Medium]]></description>
        <link>https://medium.com/@lgndryhappiness?source=rss-c08368f2ae94------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*ZJwTZCx1S9PPUTJx.png</url>
            <title>Stories by Rob Peart on Medium</title>
            <link>https://medium.com/@lgndryhappiness?source=rss-c08368f2ae94------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 06:05:29 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@lgndryhappiness/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[Automation Threatens to Make Graphic Designers Obsolete]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/aiga-eye-on-design/automation-threatens-to-make-graphic-designers-obsolete-a01579c94ae2?source=rss-c08368f2ae94------2"><img src="https://cdn-images-1.medium.com/max/1600/0*0_EdxB-EabJ7AoBf.jpg" width="1600"></a></p><p class="medium-feed-snippet">So what kind of career should you expect to have in future?</p><p class="medium-feed-link"><a href="https://medium.com/aiga-eye-on-design/automation-threatens-to-make-graphic-designers-obsolete-a01579c94ae2?source=rss-c08368f2ae94------2">Continue reading on AIGA Eye on Design »</a></p></div>]]></description>
            <link>https://medium.com/aiga-eye-on-design/automation-threatens-to-make-graphic-designers-obsolete-a01579c94ae2?source=rss-c08368f2ae94------2</link>
            <guid isPermaLink="false">https://medium.com/p/a01579c94ae2</guid>
            <category><![CDATA[work]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[future]]></category>
            <category><![CDATA[future-of-work]]></category>
            <category><![CDATA[design]]></category>
            <dc:creator><![CDATA[Rob Peart]]></dc:creator>
            <pubDate>Fri, 02 Nov 2018 07:15:00 GMT</pubDate>
            <atom:updated>2018-11-02T21:39:02.154Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Testbed is real]]></title>
            <link>https://medium.com/testbed/testbed-is-real-20c4eb125ebc?source=rss-c08368f2ae94------2</link>
            <guid isPermaLink="false">https://medium.com/p/20c4eb125ebc</guid>
            <category><![CDATA[design]]></category>
            <category><![CDATA[rapid-prototyping]]></category>
            <category><![CDATA[business]]></category>
            <category><![CDATA[prototyping]]></category>
            <dc:creator><![CDATA[Rob Peart]]></dc:creator>
            <pubDate>Fri, 28 Jul 2017 21:44:00 GMT</pubDate>
            <atom:updated>2017-07-31T07:04:26.410Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*hJXvzxqIvpupRD8u8TNcdA.png" /></figure><p><em>Today I’m pleased to announce my new venture, </em><a href="http://testbed.work/"><em>Testbed</em></a><em>. Some of you will have heard me talking about this, and I’ll have bothered a lot of you with requests for feedback and advice—thank you to all of those I have spoken with, you’ve all been very generous with your time and invaluable feedback. Here I wanted to take some time to outline some of the thinking behind the venture and some of our plans for the future. You can stay up to date via our </em><a href="https://testbed.work/"><em>mailing list</em></a><em>, </em><a href="https://www.instagram.com/wearetestbed/"><em>instagram</em></a><em> or by following our </em><a href="https://medium.com/testbed"><em>Medium publication</em></a><em>.</em></p><p>I firmly believe that creative and potentially world-changing ideas can come from anyone. Not just designers, engineers and artists, but also those working outside of those domains and on the front-line of experience. The hard parts are knowing which are the good, relevant ones and then getting those out into the world—not everyone has the tools, skills or resources to properly explore their ideas or move them onwards a step. Testbed’s mission is to help <em>anyone</em> to get their extraordinary ideas off the drawing board (or Powerpoint deck) and into reality, working with them to prototype and test them quickly and effectively.</p><p>It’s not just those without creative experience that need help. Sometimes as designers we have to work cross-discipline and need a little assistance understanding how a big idea can be expressed in ways that go beyond our training or domain knowledge. We’re here to provide that assistance.</p><p>We are currently in the process of building a workshop and knowledge-base focussed on rapid-prototyping tools and technologies that will enable us to breathe life into ideas in almost any format, from simple furniture prototypes to web-based applications to all-out physical/digital experiences. Most of our tools are built from scratch—we’re embracing the learning experience of building our own CNC, laser-cutter, workspace and more, and will document the process as we go.</p><h3><strong>Where it started</strong></h3><p>Over the past few years working in various creative and design agencies, I saw my role expand from that of a graphic designer—expected to have an opinion and a craft in expressing ideas visually—to a more multi-disciplinary designer, expected to have an expert opinion on everything from strategy to interiors to interactions. Personally I relished the challenges that the shift brought, and I have always felt that a designer <em>should </em>be multi-disciplinary by default. But it’s not always easy to embrace that shift.</p><p>The challenges come in <strong>exploring</strong>, <strong>understanding</strong> and <strong>communicating</strong> ideas that require expertise beyond your personal capabilities or experience. To do any of those three things you have to be working with the relevant tools in the relevant mediums: <strong>exploration</strong> requires first-hand and direct working with the relevant medium, not only so that the originator of the idea can begin to understand it better but also so that it can be tested for appropriateness and suitability with users; <strong>understanding</strong> of the limitations, pitfalls and benefits of the idea only come from rigorous exploration; and <strong>communication</strong> relies on a tangible anchor that all stakeholders can align their understanding against. In a situation in which only proposals have been made—instead of prototypes—one of three things can happen: you can fluke it and get a fantastic outcome that surpasses all expectations; you can let the idea die quietly as you have no idea how to take it further; or the idea fumbles its way to production, under-explored, misunderstood by yourself and stakeholders and next-to useless (or even insulting) for users. I have seen all of those three things happen at various points, and all could have been avoided by putting prototypes—and therefor users—at the centre of the design process. Testbed wants more excellent ideas to see the light of day, and less poor (or poorly executed) ideas to slip through the net. Only by realising ideas early through prototyping and subsequently testing them properly will the distinction be made.</p><h3><strong>How it developed</strong></h3><p>I have been trying to take my own advice over the past few months and have been loosely prototyping and testing my idea for Testbed. It’s been a truly eye-opening experience, and the initial thinking has become clearer and more focussed as a result.</p><p>Over the course of those discussions I’ve had to rethink who my potential customers would be, how I pitched the idea, how I framed up my services, and even re-thought the name multiple times (thanks Will).</p><p>There were also some wonderful, useful insights. One that was particularly interesting to me was the idea that most new technologies are waves to be caught or ignored—if you don’t catch the wave then don’t worry, there will be another novel technology along shortly, and you can jump on that one. This thought was surprisingly common and revealed that sometimes the extent of creative thinking with technology goes no further than a novelty, ‘we-did-it-first’ factor. There’s much more to be explored with technology as a means-to-an-end, a facilitator of experiences or part of a workflow, rather than the end or experience itself. I hope that Testbed can challenge some of the first-past-the-post mentality and help develop experiences with more direct and indirect value than sheer novelty.</p><p>Another thing that stuck with me was my friend Hector, quoting his mum: “you don’t go to art school to learn how to be an artist, you go there to learn how to think”. It’s something I think we all know, but rarely acknowledge. As designers we’re encouraged to be inventive and curious in our education, to apply our creativity in broad, boundless ways but also to be critical and open to new perspectives. Then when we enter the workplace this falls by the wayside as we apply our inventiveness within a tight discipline, on tight briefs. With Testbed we hope to make it easier to explore ideas that are outside of those tight bounds, and to bring fresh feedback and critical thinking back into the process.</p><h3><strong>Plans for the future</strong></h3><p>Currently we have our core five offerings: prototyping &amp; testing; consulting &amp; concepting; deep-dives; demo-days and workshops. We are also building our workshop and suite of tools and skills which we’d like to open up beyond our commercial offering. We plan to be able to offer short residencies for artists, makers and other curious people to use our tools and workshop to explore their own practise, and we encourage them to get in touch with proposals that can help us challenge our own thinking around design, technology and prototyping.</p><p>We’ll also be offering drop-in consulting sessions on occasional free-Fridays for anyone to run an idea past us, get advice on a project, or even just learn about our tools and our approach. We encourage you to follow our social media channels to find out when these sessions will be.</p><p>And we will also continue to update this Medium publication as a record of our progress and thinking. We hope to publish regular explorations into various technologies as foundations for creative exploration. So do keep checking in.</p><p>Consider Testbed a prototype itself: I expect the venture to morph and shift as time goes on, and I fully welcome any feedback, ideas or suggestions that will help us take it forwards.</p><p>And of course, if you need any help making your ideas real, then please, do get in touch.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=20c4eb125ebc" width="1" height="1" alt=""><hr><p><a href="https://medium.com/testbed/testbed-is-real-20c4eb125ebc">Testbed is real</a> was originally published in <a href="https://medium.com/testbed">Testbed</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Neural nets, graphic design and death metal]]></title>
            <link>https://medium.com/@lgndryhappiness/neural-nets-graphic-design-and-death-metal-40d058eb65d1?source=rss-c08368f2ae94------2</link>
            <guid isPermaLink="false">https://medium.com/p/40d058eb65d1</guid>
            <category><![CDATA[graphic-design]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[death-metal]]></category>
            <category><![CDATA[art]]></category>
            <category><![CDATA[neural-networks]]></category>
            <dc:creator><![CDATA[Rob Peart]]></dc:creator>
            <pubDate>Wed, 05 Apr 2017 15:02:38 GMT</pubDate>
            <atom:updated>2017-04-05T15:02:38.103Z</atom:updated>
            <content:encoded><![CDATA[<p><em>On Wednesday 28 March 2017, I gave a talk to the </em><a href="https://www.meetup.com/Singapore-Creative-Coding-Meetup/events/238608373/"><em>Singapore Creative Code meetup</em></a><em> about my experiments with neural nets, graphic design and death metal. Total hero </em><a href="http://www.sojamo.com/"><em>Andreas Schlegel</em></a><em> was also talking, and you can find his slides </em><a href="https://www.slideshare.net/sojamo/creative-coding-meetup-singapore-march-2017"><em>here</em></a><em>. Below is a write-up of my talk. You can see more experiments on my </em><a href="https://www.instagram.com/legendaryhappiness/"><em>instagram</em></a><em>, or follow me on </em><a href="https://twitter.com/lgndryhappiness"><em>twitter</em></a><em>.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_y1anUt_FbvJcdhj8ihEEw.png" /></figure><p>Currently I’m trying to develop some basic knowledge around how to work with neural nets and machine learning so that I might be able to undertake some deeper form of enquiry later on, or use the techniques in projects. And the best way to do that seemed to be just get my hands dirty and dive in.</p><p>Many of these explorations are totally open-ended, and not really practical. But there are a lot of sideline things that I’ve learned along the way that have been so useful. I’ll try and share these things throughout.</p><p>I’ll give you a short run down of how I got into it, then some early explorations, then getting to the training and generating of death metal logos.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*YBLI6PxP3lG0pwQxo27g7A.png" /><figcaption>A face generated by <a href="https://github.com/soumith/dcgan.torch">DCGAN</a> while looking into album covers</figcaption></figure><p>To give a bit of context, my background is in graphic design. I’m really a visual person, which is what I do for my job. I’m definitely not a skilled technician when it comes to code, but I’ve always had a keen interest in technology and I’m always looking for ways to mix the two. My interest in technology isn’t just from a visual standpoint, although that’s definitely part of it and I’m enjoying the unpredictable output as a new source of visual language; but also from a critical perspective: how much might these new technologies change my role as a designer? What about my process? What might my purpose be in the future? <a href="http://eyeondesign.aiga.org/automation-threatens-to-make-graphic-designers-obsolete/">I wrote about a few of these issues a while ago for AIGA</a>, and that spurred me to want to explore further.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*6V07JkcMmYsGvYMpVb6neA.png" /><figcaption><a href="http://www.jon.gold/2016/05/robot-design-school/">Taking the robots to design school</a>, Jon Gold</figcaption></figure><p>My interest in exploring automation and design properly started when I read this article by a designer called John Gold. It’s called ‘<a href="http://www.jon.gold/2016/05/robot-design-school/">Taking the Robots to Design School</a>’.</p><p>In it he outlines his own work in which he undertakes some analysis of commonly used typeface pairs, and algorithmically generates ideas for new typeface pairings that he may not have considered before. It’s a good illustration of how something that seems so big and scary can be applied to very tight problems and be genuinely helpful.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ujmRSrH8YCBDFsj-RupoZQ.png" /><figcaption><a href="https://medium.com/@kcimc/a-return-to-machine-learning-2de3728558eb">Image field for the “mushroom” class</a>, Kyle McDonald</figcaption></figure><p>Later I found an article by Kyle McDonald called ‘<a href="https://medium.com/@kcimc/a-return-to-machine-learning-2de3728558eb">A return to machine learning</a>’. By this point I’d been looking at <a href="https://www.coursera.org/learn/machine-learning">Andrew Ng’s Coursera</a> course and getting extremely freaked out, so to find this beautifully written and illustrated write up was invaluable. It made machine learning suddenly seem accessible to me, a halfwit designer who could barley use the terminal, let alone train NNs.</p><p>There’s a bunch of extremely helpful links. I basically went through the article methodically exploring each and every one. There’s all the stuff that you’ve all probably heard of before like deep-dream and neural style transfer, but at this point it all seemed waaaaay above my head.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Q9bgfi_kp34LJ6QSPLNrtw.png" /><figcaption><a href="http://www.jeffreythompson.org/blog/2016/03/25/torch-rnn-mac-install/">Jeffrey Thompson</a></figcaption></figure><p>One of the links introduces you to Andrej Karpathy’s <a href="https://github.com/karpathy/char-rnn">Char-RNN</a> and Justin Johnson’s<a href="https://github.com/jcjohnson/torch-rnn"> Torch-RNN</a>. Both are <a href="https://en.wikipedia.org/wiki/Recurrent_neural_network">recurrent neural networks</a> that can take a corpus of data as an input, and over the training period will begin to learn the linear structure of the data—in this case text.</p><p>By now, I didn’t know what a recurrent neural network is. And really I still only have a vague understanding at best. But it just sounded like magic, and I got excited.</p><p>On the Torch-RNN Github there was a link to an <a href="http://www.jeffreythompson.org/blog/2016/03/25/torch-rnn-mac-install/">amazing tutorial by a guy called Jeff Thompson</a> spoon-feeding you how to set everything up—Torch, CUDA, all that crap. So I did. And miraculously it worked. I can’t tell you how helpful that tutorial was. Up until now I’d been trying desperately to get Tensorflow running, reading tutorials about Theano and Caffe and just having no luck at all. To this day I am still using Torch, for the simple reason that it’s the only thing I could get up and running.</p><p>Above is some sample text after being trained on the complete works of Shakespeare by Mr Thompson. And it’s addictive, training for an hour or so, just typing out commands and getting text back that sounds like the main-man wrote it.</p><p>However, training on Shakespeare is, well, a bit boring. I wanted something more … puerile?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*9iNDSp77XRC0tYKXHbhmEQ.png" /><figcaption>Training lols</figcaption></figure><p>So I found a website with some terrrrrrible jokes, and trained it on those .</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*F484mEt63dgh-UfjPVc_7w.png" /><figcaption>RNNs have weird humour</figcaption></figure><p>Sooooo. I didn’t train it for long, and it generated the above. But I was amazed that although it’s nonsensical, there was some structure there. The call and response, even some proper words. It’s easy to see why this technique is well demonstrated by Shakespeare, given that the above sounds like some medieval standup routine. I’ve since found some killer examples of other generated text <a href="http://lewisandquark.tumblr.com/">over here</a>.</p><iframe src="https://cdn.embedly.com/widgets/media.html?type=text%2Fhtml&amp;key=d04bfffea46d4aeda930ec88cc64b87c&amp;schema=instagram&amp;url=https%3A//www.instagram.com/p/BPSpqA0DUcE/&amp;image=https%3A//i.embed.ly/1/image%3Furl%3Dhttps%253A%252F%252Fscontent.cdninstagram.com%252Ft51.2885-15%252Fs640x640%252Fe15%252F15875717_352122028506336_1226609614259224576_n.jpg%26key%3D4fce0568f2ce49e8b54624ef71a8a5bd" width="640" height="640" frameborder="0" scrolling="no"><a href="https://medium.com/media/057d1494c2caebc06c852fa7f1a2880e/href">https://medium.com/media/057d1494c2caebc06c852fa7f1a2880e/href</a></iframe><p>On my way to the death metal logos I explored a few other nets that are work mentioning.</p><p>Firstly <a href="https://pjreddie.com/darknet/">Darknet</a> by PJ Reddie, which is a great object classifier that you can either run through your terminal as a C++ thing (which is how I created the above), or now there’s <a href="https://github.com/mrzl/ofxDarknet">OFXDarknet</a> that you can work with in OpenFrameworks. It classifies still images or videos, and you can use pre-trained models on a number of different datasets. In the video version above I used a pre-trained model but linked to the wrong set of object labels, which is obviously stupid. But I found it quite unexpected and interesting.</p><iframe src="https://cdn.embedly.com/widgets/media.html?type=text%2Fhtml&amp;key=d04bfffea46d4aeda930ec88cc64b87c&amp;schema=instagram&amp;url=https%3A//www.instagram.com/p/BPaMryxjilM/&amp;image=https%3A//i.embed.ly/1/image%3Furl%3Dhttps%253A%252F%252Fscontent.cdninstagram.com%252Ft51.2885-15%252Fs640x640%252Fe15%252F16122972_1903606363202067_7981288570256097280_n.jpg%26key%3D4fce0568f2ce49e8b54624ef71a8a5bd" width="640" height="640" frameborder="0" scrolling="no"><a href="https://medium.com/media/d5629660d6f2ed312486e2bbe7990e40/href">https://medium.com/media/d5629660d6f2ed312486e2bbe7990e40/href</a></iframe><p>I followed that mistake a little bit and sent Darknet for a psychology test. This was made using the OFXDarknet addon for OpenFrameworks. Darknet likes cake.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*iloK2Vt9KPn4T_HNi011Yg.png" /><figcaption><a href="https://affinelayer.com/pixsrv/">Pix2Pix demo</a></figcaption></figure><p>Next up is <a href="https://github.com/phillipi/pix2pix">Pix2Pix</a>. You’ve probably all seen this by now on the website above, or through the wonderful work of <a href="https://twitter.com/quasimondo">Mario Klingemann</a>. It’s a lot of fun.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*TQjUDAnfminqMDllp6RGxw.png" /><figcaption>Image pairs for training</figcaption></figure><p>You train Pix2Pix on pairs of images like the one on the left. One image acts as a guide, and the other acts as a target image. The net begins to recognise the relationships between pairs, and eventually you can provide it a guide image and it will attempt to generate a target.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/496/1*-SeL7eFBBku5Sn52qQXZlA.png" /><figcaption>edges2handbags</figcaption></figure><p>As with Darknet there are some pre-trained models that can be downloaded and played with.</p><p>One in particular that took my fancy was the edges2handbags dataset. This dataset is created from a ton of images of handbags from Amazon, and trained using an algorithmically created outline as an image pair. The output examples are quite astounding.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2Foio9eBBlyzE%3Ffeature%3Doembed&amp;url=http%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3Doio9eBBlyzE&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2Foio9eBBlyzE%2Fhqdefault.jpg&amp;key=d04bfffea46d4aeda930ec88cc64b87c&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/dc0d053d58943a6e40dc0b27a5cd3c72/href">https://medium.com/media/dc0d053d58943a6e40dc0b27a5cd3c72/href</a></iframe><p>So, here’s me as handbags. I love the painterly feel of this, although it’s also quite sicky.</p><p>There are some peripheral things to mention here: the program is too large to run on my computer’s GPU. It just can’t take it, so I had to learn how to set up an Amazon GPU instance, which is not as easy as it could be. I’ve been using <a href="https://aws.amazon.com/marketplace/pp/B01LZMLK1K">this one</a> as it has the CUDA toolkit already set up and ready to go. Also, working with <a href="http://www.imagemagick.org/script/index.php">ImageMagick</a> and <a href="http://ffmpeg.org/">ffmpeg</a> made these videos much easier, and until undertaking this project I had never heard of either of them. I’m still using ImageMagick to preprocess datasets for training. It’s excellent.</p><iframe src="https://cdn.embedly.com/widgets/media.html?type=text%2Fhtml&amp;key=d04bfffea46d4aeda930ec88cc64b87c&amp;schema=instagram&amp;url=https%3A//www.instagram.com/p/BQYAeutjpQz/&amp;image=https%3A//i.embed.ly/1/image%3Furl%3Dhttps%253A%252F%252Fscontent.cdninstagram.com%252Ft51.2885-15%252Fs640x640%252Fe15%252F16583931_1348071185232368_9131147204835147776_n.jpg%26key%3D4fce0568f2ce49e8b54624ef71a8a5bd" width="640" height="640" frameborder="0" scrolling="no"><a href="https://medium.com/media/897af20ee05132dba7dbc239d1619668/href">https://medium.com/media/897af20ee05132dba7dbc239d1619668/href</a></iframe><p>I continued exploring Pix2Pix for a while, mainly because there’s a lot of room for exploration and experimentation. There’s a tight set of criteria to stick to, but that’s what makes it so usable. You can have endless fun with it. The above is a video that’s created from a dataset trained on a maps2satellite image pair taken from google maps. I created a screen recording of me zooming in to my hometown (shoutout Slough!), then used ffmpeg to split that up into individual frames. I then batch-processed the images through Pix2Pix and stitched the outputs into the video, again using ffmpeg.</p><iframe src="https://cdn.embedly.com/widgets/media.html?type=text%2Fhtml&amp;key=d04bfffea46d4aeda930ec88cc64b87c&amp;schema=instagram&amp;url=https%3A//www.instagram.com/p/BQA8h7JDRe6/&amp;image=https%3A//i.embed.ly/1/image%3Furl%3Dhttps%253A%252F%252Fscontent.cdninstagram.com%252Ft51.2885-15%252Fs640x640%252Fe15%252F16465433_1836844533271349_5872991091468271616_n.jpg%26key%3D4fce0568f2ce49e8b54624ef71a8a5bd" width="640" height="640" frameborder="0" scrolling="no"><a href="https://medium.com/media/684f5710e2d846aaea1c71723b15f240/href">https://medium.com/media/684f5710e2d846aaea1c71723b15f240/href</a></iframe><p>And of course as a designer, one can’t help but fuck about with typography. So here’s an output created by feeding letter-shaped guide images into the facades dataset. I’d like to try to find a use for these insane letterforms.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*teCXc9AlFFtHB5cOSd-4YA.png" /><figcaption>Hi m8</figcaption></figure><p>Then I found the <a href="https://github.com/soumith/dcgan.torch">DCGAN</a>. This is the one that I still love playing with. You train DCGAN on a database of images then ask it to create new ones from the model. It’s an adversarial network, which means that you actually train two nets that kind-of duke-it-out during the training process and try to slip each other up. You can read about it on the Github and associated paper.</p><p>One of the models you can download has been pre-trained on 10000 images of celebrities. Which is fun. The great thing is that you can generate the new images in multiple ways. You can generate single images like the one above.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/922/1*nQ2b1Y8nj5TEy-moUOLnEQ.png" /><figcaption>Hi m8s</figcaption></figure><p>Or if you’re short of friends you can generate a tapestry of new ones, like this. Silk scarf, anyone?</p><iframe src="https://cdn.embedly.com/widgets/media.html?type=text%2Fhtml&amp;key=d04bfffea46d4aeda930ec88cc64b87c&amp;schema=instagram&amp;url=https%3A//www.instagram.com/p/BQLNxxFD6Bk/&amp;image=https%3A//i.embed.ly/1/image%3Furl%3Dhttps%253A%252F%252Fscontent.cdninstagram.com%252Ft51.2885-15%252Fe15%252F16464104_261667220933540_3690621197944356864_n.jpg%26key%3D4fce0568f2ce49e8b54624ef71a8a5bd" width="640" height="640" frameborder="0" scrolling="no"><a href="https://medium.com/media/e908c9dfdee1385e0708de43710b708a/href">https://medium.com/media/e908c9dfdee1385e0708de43710b708a/href</a></iframe><p>My favourite way to work with it is to have the net create a series of images that ‘step’ between two generated outputs, then stitch them into videos like the above.</p><p>By now I was getting more comfortable with the way it all works and I was about ready to try and train my own network.</p><iframe src="https://cdn.embedly.com/widgets/media.html?url=http%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DGAJByueLRTo&amp;src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FGAJByueLRTo%3Ffeature%3Doembed&amp;type=text%2Fhtml&amp;key=d04bfffea46d4aeda930ec88cc64b87c&amp;schema=youtube" width="640" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/d4ec46925d693849111d51fffdc14a02/href">https://medium.com/media/d4ec46925d693849111d51fffdc14a02/href</a></iframe><p>I’m a designer, and I’m also lazy. I found this dataset of logos. Really horrific logos. It’s clipart really. But I wanted to see what happened if I trained DCGAN on these and tried to get it to output something.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*lQQLDY4qwK1lYKYDHlhtSw.png" /></figure><p>I trained overnight, it did (I think) around 2000 iterations in 7 hours. The great thing is that you can watch the output as it’s training, to check that everything is running smoothly. On the left you can see some of the outputs while training. Notice that it’s already recognising some visual patterns, like circular shapes in the centre of the frame, or bits of typography around the edges of the logos.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/906/1*CkhSOlPq04_yZxOIhT4yjA.png" /><figcaption>Logos!</figcaption></figure><p>OK, so it’s not 100% amazing, but this blew me away. It’s trying to put text in the centre of circles, and has a reasonable idea of form. There are arcs and swooshes, roundels and triangles. I was amazed by this.</p><p>I started thinking again about Shakespeare, and how my RNN jokes had sounded like a fat monk from 1523 trying to be a comedian. I realised that the Shakespeare dataset seems to be reasonably sympathetic to the mistakes in generated output—even stuff that doesn’t go in as Shakespeare comes out sounding Ye Olde Worlde, if you train it badly enough (which I did). So I started thinking about similar datasets I could use with DCGAN …</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*P-4u2DVIV0gYyPInREnYcg.png" /><figcaption>METAL</figcaption></figure><p>I don’t know if you’re aware of death metal logos, but they take as a point-of-pride their illegibility. Here’s some nice ones. I scraped 500 of these from the bottom of the Pinterest barrel. There’s something interesting in having an artificial network generate death metal logos. It’s probably one of the more esoteric types of music around, which in my mind makes it distinctively human—you have to appreciate it not only from an aural perspective, but also from a cultural one. Machines obviously can’t do this, but still it seemed like one of the more suitable kinds of visual to attempt imitating. Perhaps it’s because of the levels of complexity that we perceive neural nets as having: we can’t easily generate these logos procedurally, so maybe a more ‘organic’ method like a neural net is more suitable?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/902/1*hHPio2r4JsMEvX_BgSVuNw.png" /><figcaption>Mid way through training</figcaption></figure><p>Midway through the training and our friend DCGAN is already starting to get familiar with the forms of the logos. I figured that if I just slammed it through thousands more iterations then something impressive would come out of it.</p><p>I was wrong.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*kT4Ic4soUPaA9PBr-mUIcA.png" /></figure><p>After about 2000 iterations, everything got very repetitive. My dataset is relatively tiny, and I’m guessing that the net is learning patterns that are frequently repeated and just outputting those. I found that going back to a checkpoint at around 1400 iterations was giving me the best output. You can see on the right here some very lo-res logos. DCGAN is set up to train images 64px square. You can make adjustments to the code, and get this up to 128 or 256 but encounter problems training. This was solved by adding in a white-noise layer to the architecture of the net. <a href="https://github.com/soumith/dcgan.torch/issues/2">There’s a good thread about it here</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*V3zuiDuplY0BNrFBKfuvBg.png" /></figure><p>I think the results speak for themselves really. On the left you have two of the better logos I was able to generate, vs two fantastic death-metal logos on the right. No-one is going out of business soon, especially not <a href="https://www.facebook.com/christophe.szpajdel?fref=nf">Christophe Szpajdel</a>.</p><p>My next step is to look into audioGANs and see if I can generate some death metal music. Then I’ll have a metal night. And you’re all invited.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*aYDqD3WncH20O1txuQTgiQ.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=40d058eb65d1" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Why Design is Not Problem Solving + Design Thinking Isn’t Always the Answer]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@lgndryhappiness/why-design-is-not-problem-solving-design-thinking-isnt-always-the-answer-be1d3c7762c5?source=rss-c08368f2ae94------2"><img src="https://cdn-images-1.medium.com/max/2000/0*OAoI_wAhsZBdg4Ls.jpg" width="2000"></a></p><p class="medium-feed-snippet">I wrote this a while back for AIGA Eye on Design. Link to original here. I&#x2019;m not too sure how I feel about it now&#x200A;&#x2014;&#x200A;it&#x2019;s overly simplistic&#x2026;</p><p class="medium-feed-link"><a href="https://medium.com/@lgndryhappiness/why-design-is-not-problem-solving-design-thinking-isnt-always-the-answer-be1d3c7762c5?source=rss-c08368f2ae94------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@lgndryhappiness/why-design-is-not-problem-solving-design-thinking-isnt-always-the-answer-be1d3c7762c5?source=rss-c08368f2ae94------2</link>
            <guid isPermaLink="false">https://medium.com/p/be1d3c7762c5</guid>
            <category><![CDATA[design]]></category>
            <dc:creator><![CDATA[Rob Peart]]></dc:creator>
            <pubDate>Thu, 19 Jan 2017 12:15:33 GMT</pubDate>
            <atom:updated>2019-11-29T14:38:32.337Z</atom:updated>
        </item>
    </channel>
</rss>