[{"data":1,"prerenderedAt":493},["ShallowReactive",2],{"footer-settings":3,"post-shell:how-artists-can-set-up-their-own-neural-network-part-4-901ef71078bb":22,"post-body:how-artists-can-set-up-their-own-neural-network-part-4-901ef71078bb":61},{"heading":4,"body":5,"links":6,"note":21},"Bottom line, still up front.","A small footer for global links, contact paths, and project context.",[7,10,13,16,19],{"label":8,"url":9},"About","http:\u002F\u002Fmy-website.localhost\u002Fabout#now",{"label":11,"url":12},"Case Studies","http:\u002F\u002Fmy-website.localhost\u002F#selected-work",{"label":14,"url":15},"Side Projects","http:\u002F\u002Fmy-website.localhost\u002Fside-projects",{"label":17,"url":18},"Writing","http:\u002F\u002Fmy-website.localhost\u002Fwriting",{"label":20,"url":9},"Now","",{"id":23,"slug":24,"date":25,"author":26,"title":29,"excerpt":30,"canonicalUrl":31,"featuredImage":32,"authorName":28,"featuredMedia":33,"blocks":60},"cG9zdDo2MTM=","how-artists-can-set-up-their-own-neural-network-part-4-901ef71078bb","April 6, 2023",{"node":27},{"name":28},"Aslan French","How Artists Can Set Up Their Own Neural Network — part 4 —","How Artists Can Set Up Their Own Neural Network — part 4 — This post was originally published in 2018\u002F2019 on my old website. I have pared down my personal website a bunch, and now am using Medium to do all…","https:\u002F\u002Faslanfrench.medium.com\u002Fhow-artists-can-set-up-their-own-neural-network-part-4-901ef71078bb",{"node":33},{"id":34,"sourceUrl":35,"srcSet":36,"sizes":37,"altText":21,"mediaDetails":38},"cG9zdDo2MTQ=","https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2026\u002F05\u002F1CKl7IBPVsMILTGgq6MWdPQ.png","https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2026\u002F05\u002F1CKl7IBPVsMILTGgq6MWdPQ-300x300.png 300w, https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2026\u002F05\u002F1CKl7IBPVsMILTGgq6MWdPQ-150x150.png 150w, https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2026\u002F05\u002F1CKl7IBPVsMILTGgq6MWdPQ-768x768.png 768w, https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2026\u002F05\u002F1CKl7IBPVsMILTGgq6MWdPQ.png 1024w","(max-width: 300px) 100vw, 300px",{"width":39,"height":39,"sizes":40},1024,[41,48,54],{"name":42,"sourceUrl":43,"width":44,"height":44,"mimeType":45,"file":46,"fileSize":47},"medium","https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2026\u002F05\u002F1CKl7IBPVsMILTGgq6MWdPQ-300x300.png","300","image\u002Fpng","1CKl7IBPVsMILTGgq6MWdPQ-300x300.png",200857,{"name":49,"sourceUrl":50,"width":51,"height":51,"mimeType":45,"file":52,"fileSize":53},"thumbnail","https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2026\u002F05\u002F1CKl7IBPVsMILTGgq6MWdPQ-150x150.png","150","1CKl7IBPVsMILTGgq6MWdPQ-150x150.png",57279,{"name":55,"sourceUrl":56,"width":57,"height":57,"mimeType":45,"file":58,"fileSize":59},"medium_large","https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2026\u002F05\u002F1CKl7IBPVsMILTGgq6MWdPQ-768x768.png","768","1CKl7IBPVsMILTGgq6MWdPQ-768x768.png",997686,[],[62,68,72,77,81,85,90,94,98,102,106,110,114,118,122,127,131,136,140,144,148,152,156,160,164,168,172,176,180,184,188,192,196,200,204,208,212,216,220,224,228,232,236,240,244,248,252,256,260,264,268,272,276,280,284,288,292,296,300,304,308,312,316,320,324,329,333,337,341,345,349,353,357,361,365,369,373,377,381,385,389,393,397,401,405,409,414,419,423,427,431,435,439,443,447,451,455,459,463,467,471,475,479,484,489],{"name":63,"clientId":64,"parentClientId":65,"renderedHtml":66,"attributes":67},"core\u002Fparagraph","6a8e6cd317ee4",null,"\n\u003Cp class=\"wp-block-paragraph\">This post was originally published in 2018\u002F2019 on my old website. I have pared down my personal website a bunch, and now am using Medium to do all my blogging. Additionally the techniques for setting up a neural network that I explain in part \u003Ca href=\"https:\u002F\u002Fmedium.com\u002Fhackernoon\u002Fhow-artists-can-set-up-their-own-neural-network-part-1-installation-5dddac1ca331\">1\u003C\u002Fa>, \u003Ca href=\"https:\u002F\u002Fmedium.com\u002Fhackernoon\u002Ftutorial-for-artists-on-how-to-use-a-neural-network-part-2-4e94e1d2cbe9\">2\u003C\u002Fa>, and \u003Ca href=\"https:\u002F\u002Fmedium.com\u002Fhackernoon\u002Fhow-artists-can-set-up-their-own-neural-network-part-3-image-generation-2e7679f6959c\">3\u003C\u002Fa> is not really relevant anymore, and these days I use Google Colab rather than run things locally. I published some of the Google Colab scripts I use now \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fjcklpe\u002Fneural-network-art\">here\u003C\u002Fa>, and \u003Ca href=\"https:\u002F\u002Fcolab.research.google.com\u002Fdrive\u002F1YEeWvuCreoIAFnhGRYQTXtF2Tlj0UCiR?usp=sharing\">here\u003C\u002Fa>.\u003C\u002Fp>\n",{},{"name":63,"clientId":69,"parentClientId":65,"renderedHtml":70,"attributes":71},"6a8e6cd317f78","\n\u003Cp class=\"wp-block-paragraph\">This fourth part though was really more focused on the manual art making process and some of my theory on how I approach it.\u003C\u002Fp>\n",{},{"name":73,"clientId":74,"parentClientId":65,"renderedHtml":75,"attributes":76},"core\u002Fseparator","6a8e6cd317fc4","\n\u003Chr class=\"wp-block-separator has-alpha-channel-opacity is-style-wide\"\u002F>\n",{},{"name":63,"clientId":78,"parentClientId":65,"renderedHtml":79,"attributes":80},"6a8e6cd317fc7","\n\u003Cp class=\"wp-block-paragraph\">Time to start getting your hands dirty with some paint. Digital paint at least.\u003C\u002Fp>\n",{},{"name":63,"clientId":82,"parentClientId":65,"renderedHtml":83,"attributes":84},"6a8e6cd317fc8","\n\u003Cp class=\"wp-block-paragraph\">I’m not going to teach you how to do all the things that are required for digital painting. That’s a whole field in itself, but I would like to show my process for working with style transfer neural networks and offer some tips on things that I think help in the process.\u003C\u002Fp>\n",{},{"name":86,"clientId":87,"parentClientId":65,"renderedHtml":88,"attributes":89},"core\u002Fheading","6a8e6cd317fc9","\n\u003Ch2 class=\"wp-block-heading\">Image Generation\u003C\u002Fh2>\n",{},{"name":63,"clientId":91,"parentClientId":65,"renderedHtml":92,"attributes":93},"6a8e6cd317fca","\n\u003Cp class=\"wp-block-paragraph\">I spoke a little bit in the last post about image generation batch scripting. You can go check \u003Ca href=\"https:\u002F\u002Fmedium.com\u002Fhackernoon\u002Fhow-artists-can-set-up-their-own-neural-network-part-3-image-generation-2e7679f6959c\">that bit\u003C\u002Fa> out, but I’ll just skip over that and talk about this aspect in a more conceptual and practical level.\u003C\u002Fp>\n",{},{"name":63,"clientId":95,"parentClientId":65,"renderedHtml":96,"attributes":97},"6a8e6cd317fcb","\n\u003Cp class=\"wp-block-paragraph\">Image generation is all about curation. It’s a process of throwing a lot of stuff at the wall and then seeing what sticks. Because it’s trivially easy to generate images due to automation and space, I often times will run the image through 100 different “source-styles”. Generating one image takes maybe 5 minutes but if you let the system run overnight you can easily get all that knocked out in around 8 hours. It works while you sleep.\u003C\u002Fp>\n",{},{"name":86,"clientId":99,"parentClientId":65,"renderedHtml":100,"attributes":101},"6a8e6cd317fcc","\n\u003Ch2 class=\"wp-block-heading\">Picking your style source images.\u003C\u002Fh2>\n",{},{"name":86,"clientId":103,"parentClientId":65,"renderedHtml":104,"attributes":105},"6a8e6cd317fcd","\n\u003Ch3 class=\"wp-block-heading\">Keep detail low\u003C\u002Fh3>\n",{},{"name":63,"clientId":107,"parentClientId":65,"renderedHtml":108,"attributes":109},"6a8e6cd317fce","\n\u003Cp class=\"wp-block-paragraph\">You want to think of of your style source images like wallpaper, or texture. Its like a landscape. It’s got to be broad. It can’t be too detailed. One of the reasons I say this is because you’re going to get a lot of details blurred out through the downsizing process. I run my neural network on my home cloud server. It has a 970. It’s a bit dated but I also think you’d probably have similar resolution issues with newer cards. If you increase the size of an image without distorting it’s ratio you have to size up both width and height which means for ever pixel of height you had you also multiplied by the width. So a card would probably have to be multiple times more powerful than the 970 to increase the resolution up much more. That’s why I enlarge the images after the fact because the only way to get usable images is to downsize them to longest side being 500px. The resulting images can be scaled up by Waifu2x but the image that goes in has to be pretty small first.\u003C\u002Fp>\n",{},{"name":63,"clientId":111,"parentClientId":65,"renderedHtml":112,"attributes":113},"6a8e6cd317fcf","\n\u003Cp class=\"wp-block-paragraph\">That process loses detail. So broad images are good. I think it’s good to get an image that doesn’t feel like it is too strong any particular subject in it’s composition.\u003C\u002Fp>\n",{},{"name":86,"clientId":115,"parentClientId":65,"renderedHtml":116,"attributes":117},"6a8e6cd317fd0","\n\u003Ch2 class=\"wp-block-heading\">Balance of high and low contrast areas\u003C\u002Fh2>\n",{},{"name":63,"clientId":119,"parentClientId":65,"renderedHtml":120,"attributes":121},"6a8e6cd317fd1","\n\u003Cp class=\"wp-block-paragraph\">You want something that has kind of a lower contrast low noise interrupted by areas of sharp transition in hue, value, and saturation.\u003C\u002Fp>\n",{},{"name":123,"clientId":124,"parentClientId":65,"renderedHtml":125,"attributes":126},"core\u002Fimage","6a8e6cd317fd2","\u003Cdiv class=\"wp-block-image\">\n\u003Cfigure class=\"alignleft size-full\">\u003Cimg src=\"https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2023\u002F04\u002Fmedium-6e747e62b1aa654ac067d024fab4e605.jpg\" alt=\"\" class=\"wp-image-694\"\u002F>\u003C\u002Ffigure>\n\u003C\u002Fdiv>",{},{"name":63,"clientId":128,"parentClientId":65,"renderedHtml":129,"attributes":130},"6a8e6cd317fd3","\n\u003Cp class=\"wp-block-paragraph\">Color can be mapped into 3 dimensions. Hue (the color like red or green), saturation (how much color versus grey there is), and value (the darkness or brightness of the color).\u003Ca href=\"http:\u002F\u002Fsmallhadroncollider-labs.github.io\u002Frgb-hsl\u002F\"> Here’s\u003C\u002Fa> a 3D visualization toy\u002Ftool you can play around with to understand it better, or a\u003Ca href=\"http:\u002F\u002Fhslpicker.com\u002F#74cda7\"> 2D\u003C\u002Fa> one if you prefer. This\u003Ca href=\"https:\u002F\u002Fprogrammingdesignsystems.com\u002Fcolor\u002Fcolor-models-and-color-spaces\u002Findex.html\"> article\u003C\u002Fa> also has some great 3d visualization tools you can play with and gives a more in depth explanation of the different ways in which color variables can be mapped in 3D space.\u003C\u002Fp>\n",{},{"name":132,"clientId":133,"parentClientId":65,"renderedHtml":134,"attributes":135},"core\u002Fgroup","6a8e6cd317fd4","\n\u003Cdiv class=\"wp-block-group is-nowrap is-layout-flex wp-container-core-group-is-layout-8f761849 wp-block-group-is-layout-flex\">\n\u003Cp class=\"wp-block-paragraph\">High contrast areas are going to be ones in which colors which are next to each other have higher variance along one of these axis.\u003C\u002Fp>\n\u003C\u002Fdiv>\n",{},{"name":63,"clientId":137,"parentClientId":133,"renderedHtml":138,"attributes":139},"6a8e6cd317fd5","\n\u003Cp class=\"wp-block-paragraph\">High contrast areas are going to be ones in which colors which are next to each other have higher variance along one of these axis.\u003C\u002Fp>\n",{},{"name":86,"clientId":141,"parentClientId":65,"renderedHtml":142,"attributes":143},"6a8e6cd317fdb","\n\u003Ch3 class=\"wp-block-heading\">Value\u003C\u002Fh3>\n",{},{"name":123,"clientId":145,"parentClientId":65,"renderedHtml":146,"attributes":147},"6a8e6cd317fdc","\u003Cdiv class=\"wp-block-image\">\n\u003Cfigure class=\"alignright size-full\">\u003Cimg src=\"https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2023\u002F04\u002Fmedium-1dbd2f044fb46e48ab894fe0a41b431c.jpg\" alt=\"\" class=\"wp-image-695\"\u002F>\u003Cfigcaption class=\"wp-element-caption\">\u003Ca href=\"https:\u002F\u002Fpixabay.com\u002Fphotos\u002Fmoss-hdr-wood-forest-green-log-1551891\u002F\">credit to photographer\u003C\u002Fa>\u003C\u002Ffigcaption>\u003C\u002Ffigure>\n\u003C\u002Fdiv>",{},{"name":63,"clientId":149,"parentClientId":65,"renderedHtml":150,"attributes":151},"6a8e6cd317fdd","\n\u003Cp class=\"wp-block-paragraph\">Value contrast is by far the most important, probably because the majority of your eyesight comes from rods rather than cones. Rods are more numerous and sensitive, but don’t detect color​​\u003Csup data-fn=\"bfb2f344-67dd-47a2-b017-8de6ddc9f325\" class=\"fn\">\u003Ca href=\"#bfb2f344-67dd-47a2-b017-8de6ddc9f325\" id=\"bfb2f344-67dd-47a2-b017-8de6ddc9f325-link\">1\u003C\u002Fa>\u003C\u002Fsup>, while cones detect color and value. Value contrast can most easily be seen in HDR photography, where the camera takes a wider range of dark and light values through multiple exposures and compresses them to retain a greater amount of value contrast information.\u003C\u002Fp>\n",{},{"name":86,"clientId":153,"parentClientId":65,"renderedHtml":154,"attributes":155},"6a8e6cd317fde","\n\u003Ch3 class=\"wp-block-heading\">Hue\u003C\u002Fh3>\n",{},{"name":63,"clientId":157,"parentClientId":65,"renderedHtml":158,"attributes":159},"6a8e6cd317fdf","\n\u003Cp class=\"wp-block-paragraph\">Hue is simply what you think of as color normally. Red, blue, green etc. The contrast between hues is generally greatest between\u003Ca href=\"https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FComplementary_colors\"> complementary colors\u003C\u002Fa>. Orange and blue provide good contrast against one another, which is why it used to be\u003Ca href=\"https:\u002F\u002Fwww.thewrap.com\u002Fwhy-movie-posters-all-look-same-75846\u002F\"> so common in film posters\u003C\u002Fa>.\u003C\u002Fp>\n",{},{"name":123,"clientId":161,"parentClientId":65,"renderedHtml":162,"attributes":163},"6a8e6cd317fe0","\u003Cdiv class=\"wp-block-image\">\n\u003Cfigure class=\"alignleft size-full\">\u003Cimg src=\"https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2023\u002F04\u002Fmedium-b3b566a7b3787d9a47fdbf33b0c5c033.jpg\" alt=\"\" class=\"wp-image-696\"\u002F>\u003C\u002Ffigure>\n\u003C\u002Fdiv>",{},{"name":63,"clientId":165,"parentClientId":65,"renderedHtml":166,"attributes":167},"6a8e6cd317fe1","\n\u003Cp class=\"wp-block-paragraph\">ah the days of pre-Avengers blockbusters\u003C\u002Fp>\n",{},{"name":63,"clientId":169,"parentClientId":65,"renderedHtml":170,"attributes":171},"6a8e6cd317fe2","\n\u003Cp class=\"wp-block-paragraph\">Your eyes do not perceive all colors equally well. This is referred to as relative\u003Ca href=\"http:\u002F\u002Fwww.workwithcolor.com\u002Fcolor-luminance-2233.htm\"> color luminosity\u003C\u002Fa>. Colors like red, and blue are perceived as “darker” due to them being on the edges of our color vision. This is because our lowest frequency perceptions of light start at red, and go up to blue. Purple is a color we see due to interference between the lowest frequency receptors (red) and the highest frequency receptors(blue). In some sense purple is an illusion. Though I think the interference might be inherent even if we were capable of seeing beyond blue into ultraviolet light because ultraviolet light is roughly 800THZ, making it basically an “octave” higher than red at 400THZ​​. (also neat article on some of the history of overlap between color theory and music theory\u003Ca href=\"https:\u002F\u002Froelhollander.eu\u002Fen\u002Ftuning-frequency\u002Fsound-light-colour\u002F\"> here\u003C\u002Fa>).\u003C\u002Fp>\n",{},{"name":63,"clientId":173,"parentClientId":65,"renderedHtml":174,"attributes":175},"6a8e6cd317fe3","\n\u003Cp class=\"wp-block-paragraph\">Besides aiming for complementary colors you can also try to pair colors with high and low color luminosity. Red and Blue have a certain amount of contrast because they’re pretty far apart from each other on the spectrum but they’re both perceived with low color luminescence which means they can get sort of muddy. (think how the American flag would look without the white bars and stripes to break things up.)\u003C\u002Fp>\n",{},{"name":86,"clientId":177,"parentClientId":65,"renderedHtml":178,"attributes":179},"6a8e6cd317fe4","\n\u003Ch3 class=\"wp-block-heading\">Saturation\u003C\u002Fh3>\n",{},{"name":63,"clientId":181,"parentClientId":65,"renderedHtml":182,"attributes":183},"6a8e6cd317fe5","\n\u003Cp class=\"wp-block-paragraph\">Saturation contrast is generally less used but it’s quite striking when it is, such as in Sin City or Schindler’s list. Saturation contrast is pretty significant but comes with the problem of having trouble mapping semantically to your source content images. If you have an image that is gray with blue flowers and you try to apply that as a style to a portrait then you might end up with a face with a bunch of random blue splotches on it because the neural network is trying to follow differences in form and edges so it doesn’t really understand foreground\u002Fbackground and object differentiation.\u003C\u002Fp>\n",{},{"name":123,"clientId":185,"parentClientId":65,"renderedHtml":186,"attributes":187},"6a8e6cd317fe6","\n\u003Cfigure class=\"wp-block-image size-full\">\u003Cimg src=\"https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2023\u002F04\u002Fmedium-0800244b9f112051f50c213f3d35ab7a.jpg\" alt=\"\" class=\"wp-image-697\"\u002F>\u003C\u002Ffigure>\n",{},{"name":63,"clientId":189,"parentClientId":65,"renderedHtml":190,"attributes":191},"6a8e6cd317fe7","\n\u003Cp class=\"wp-block-paragraph\">A good reference point for what kinds of images you should be looking for can be best described in terms of color blocking.\u003Ca href=\"https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FColor-blocking\"> Color blocking\u003C\u002Fa> is a thing done in fashion inspired by the work of\u003Ca href=\"https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FColor-blocking\"> Piet Mondrian\u003C\u002Fa>. A strongly color blocked scene is going to areas of flatness butted up against other areas of flatness which also have high contrast to their neighbors.\u003C\u002Fp>\n",{},{"name":86,"clientId":193,"parentClientId":65,"renderedHtml":194,"attributes":195},"6a8e6cd317fe8","\n\u003Ch3 class=\"wp-block-heading\">Some contrast but not too much\u003C\u002Fh3>\n",{},{"name":63,"clientId":197,"parentClientId":65,"renderedHtml":198,"attributes":199},"6a8e6cd317fe9","\n\u003Cp class=\"wp-block-paragraph\">It’s hard for me to reliably say how much or which kinds of contrasts are best because there are so many additional variables, but I think you can look at two extremes.\u003C\u002Fp>\n",{},{"name":123,"clientId":201,"parentClientId":65,"renderedHtml":202,"attributes":203},"6a8e6cd317fea","\u003Cdiv class=\"wp-block-image\">\n\u003Cfigure class=\"alignright size-full\">\u003Cimg src=\"https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2026\u002F05\u002F1CKl7IBPVsMILTGgq6MWdPQ.png\" alt=\"\" class=\"wp-image-614\"\u002F>\u003C\u002Ffigure>\n\u003C\u002Fdiv>",{},{"name":63,"clientId":205,"parentClientId":65,"renderedHtml":206,"attributes":207},"6a8e6cd317feb","\n\u003Cp class=\"wp-block-paragraph\">Highly “patterned” images with extreme and distinct contrasts between areas can overwhelm a content image, such as in this Lisa Frank tiger pattern below.\u003C\u002Fp>\n",{},{"name":63,"clientId":209,"parentClientId":65,"renderedHtml":210,"attributes":211},"6a8e6cd317fec","\n\u003Cp class=\"wp-block-paragraph\">Insufficient contrast (or contrast which gets averaged out due to it being small in scale, such as in noise) can give insufficient texture information for the neural network to properly transfer style. (this is often true for images which are photographic in nature, as reality tends to be a lot more gradient than abstracted illustration).\u003C\u002Fp>\n",{},{"name":63,"clientId":213,"parentClientId":65,"renderedHtml":214,"attributes":215},"6a8e6cd317fed","\n\u003Cp class=\"wp-block-paragraph\">That is not to say that these kinds of images are not still useful! They can generate very interesting textures or color fields which can then be brought back into the final painting later on in light touches.\u003C\u002Fp>\n",{},{"name":123,"clientId":217,"parentClientId":65,"renderedHtml":218,"attributes":219},"6a8e6cd317fee","\u003Cdiv class=\"wp-block-image\">\n\u003Cfigure class=\"alignleft size-full\">\u003Cimg src=\"https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2023\u002F04\u002Fmedium-abf1fd377eafe6f7249f9a2635ddfae3.png\" alt=\"\" class=\"wp-image-698\"\u002F>\u003C\u002Ffigure>\n\u003C\u002Fdiv>",{},{"name":63,"clientId":221,"parentClientId":65,"renderedHtml":222,"attributes":223},"6a8e6cd317fef","\n\u003Cp class=\"wp-block-paragraph\">To understand how this works I think you have to actually look at what the neural network project says in their paper on the subject. It’s a neural network trained to recognize texture, and it’s being tasked with successively transforming the image to resemble the style source image. It’s not transferring the essence of the art​\u003Csup data-fn=\"5fd6cd63-cc21-4290-9955-fa3ca565485b\" class=\"fn\">\u003Ca href=\"#5fd6cd63-cc21-4290-9955-fa3ca565485b\" id=\"5fd6cd63-cc21-4290-9955-fa3ca565485b-link\">2\u003C\u002Fa>\u003C\u002Fsup>.\u003C\u002Fp>\n",{},{"name":63,"clientId":225,"parentClientId":65,"renderedHtml":226,"attributes":227},"6a8e6cd317ff0","\n\u003Cp class=\"wp-block-paragraph\">In some ways I think it’s helpful to think of the texture of the source style image as being “wrapped around” the source content image.\u003C\u002Fp>\n",{},{"name":63,"clientId":229,"parentClientId":65,"renderedHtml":230,"attributes":231},"6a8e6cd317ff1","\n\u003Cp class=\"wp-block-paragraph\">This is why I think impressionism tends to be such a popular choice for these neural networks. Impressionist painters tend to offer you a nice rhythm of areas of noise and areas of emptiness.\u003C\u002Fp>\n",{},{"name":63,"clientId":233,"parentClientId":65,"renderedHtml":234,"attributes":235},"6a8e6cd317ff2","\n\u003Cp class=\"wp-block-paragraph\">quick aside\u003C\u002Fp>\n",{},{"name":63,"clientId":237,"parentClientId":65,"renderedHtml":238,"attributes":239},"6a8e6cd317ff3","\n\u003Cp class=\"wp-block-paragraph\">I’m not entirely sure about this because I don’t have the necessary computer science and math background to really say, but I have a strong instinct that the reason why this kind of contrast is useful is specifically because of how edge detection works. When I was in school I did film stuff. In order to make up for my bad focus in a shot I would apply an unsharp mask.\u003Ca href=\"https:\u002F\u002Fwww.cambridgeincolour.com\u002Ftutorials\u002Funsharp-mask.htm\"> Here’s a good article\u003C\u002Fa> on the subject. Essentially an unsharp mask sharpens an image by making dark pixels nearly light pixels darker, and making light pixels near dark pixels lighter, while leaving dark pixels near dark pixels and light pixels near light pixels pretty much alone.\u003C\u002Fp>\n",{},{"name":63,"clientId":241,"parentClientId":65,"renderedHtml":242,"attributes":243},"6a8e6cd317ff4","\n\u003Cp class=\"wp-block-paragraph\">This contextually dependent transformation of pixels according to their surrounding pixels reminds me a bit of my intuition about the kinds of patterns which work best source style images.\u003C\u002Fp>\n",{},{"name":63,"clientId":245,"parentClientId":65,"renderedHtml":246,"attributes":247},"6a8e6cd317ff5","\n\u003Cp class=\"wp-block-paragraph\">The unsharp mask is called a kind of “high pass filter”, as is the\u003Ca href=\"https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FSobel_operator\"> Sobel Filter\u003C\u002Fa> which is commonly used in edge detection algorithms. In reading through some of the DeepStyle neural network articles I’m pretty sure I saw some stuff on using edge detection stuff with some of the nodes, so I think the essential intuition I have about what types of images work best comes down to the fact that images whose edge detections are easily highlighted by a high pass filter are going to be the ones which work best because the neural networks are in some way specifically leveraged on edge detection technology.\u003C\u002Fp>\n",{},{"name":86,"clientId":249,"parentClientId":65,"renderedHtml":250,"attributes":251},"6a8e6cd317ff6","\n\u003Ch2 class=\"wp-block-heading\">Picking your source content images\u003C\u002Fh2>\n",{},{"name":63,"clientId":253,"parentClientId":65,"renderedHtml":254,"attributes":255},"6a8e6cd318070","\n\u003Cp class=\"wp-block-paragraph\">It’s a good idea to have a strong single subject for source content images. This is again due to resolution issues. Landscape like images are good for source style images because they are broad, won’t overly bias the style being transferred too much, and don’t lose detail as much clarity when their resolution is low. For these same reasons landscape type images are poor content images. The neural network has trouble “wrapping” the style around the content if the content doesn’t have a strong composition. Portraits, still lives, things which have a singular subject give more clues for the neural network to wrap around. Landscapes mostly work as source content images when there’s exceptionally sharp blocking and I don’t see that as often other than maybe at the horizon line. When you use a overly broad sweep of an image, what you get back will more resemble noise.\u003C\u002Fp>\n",{},{"name":86,"clientId":257,"parentClientId":65,"renderedHtml":258,"attributes":259},"6a8e6cd318076","\n\u003Ch2 class=\"wp-block-heading\">Dialogue\u003C\u002Fh2>\n",{},{"name":63,"clientId":261,"parentClientId":65,"renderedHtml":262,"attributes":263},"6a8e6cd318077","\n\u003Cp class=\"wp-block-paragraph\">I think it’s important to explain one of the things I find most fascinating about this artistic process. Neural networks are algorithmic in a sense and performed using a computer doing math, but it’s completely unlike any\u003Ca href=\"https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FGenerative_art\"> generative art\u003C\u002Fa> we’ve ever had before. More traditional generative work is a human artist putting in data to a box\u003Csup data-fn=\"3ba9ec09-18fc-4ef8-9c94-24b27b8ae85b\" class=\"fn\">\u003Ca href=\"#3ba9ec09-18fc-4ef8-9c94-24b27b8ae85b\" id=\"3ba9ec09-18fc-4ef8-9c94-24b27b8ae85b-link\">3\u003C\u002Fa>\u003C\u002Fsup> which then spits some kind of result. It’s one direction. Using neural networks as I do is more like a two way\u003Ca href=\"https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FDialectic\"> classical dialectic\u003C\u002Fa> than a one way transformation.\u003C\u002Fp>\n",{},{"name":63,"clientId":265,"parentClientId":65,"renderedHtml":266,"attributes":267},"6a8e6cd318079","\n\u003Cp class=\"wp-block-paragraph\">Image generation through the batch process can result in unpredictable results This is a powerful creative tool because it can inspire new ideas\u003C\u002Fp>\n",{},{"name":123,"clientId":269,"parentClientId":65,"renderedHtml":270,"attributes":271},"6a8e6cd31807a","\u003Cdiv class=\"wp-block-image\">\n\u003Cfigure class=\"alignright size-full\">\u003Cimg src=\"https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2023\u002F04\u002Fmedium-e4e49079987a3bc07c5362ddbea3bf21.jpg\" alt=\"\" class=\"wp-image-699\"\u002F>\u003Cfigcaption class=\"wp-element-caption\">\u003Ca href=\"http:\u002F\u002Fwww.tate.org.uk\u002Fart\u002Fwork\u002FP78459\">Exquisite Corpse 2000 Jake Chapman and Dinos Chapman born 1966, born 1962 Purchased 2000\u003C\u002Fa>\u003C\u002Ffigcaption>\u003C\u002Ffigure>\n\u003C\u002Fdiv>",{},{"name":63,"clientId":273,"parentClientId":65,"renderedHtml":274,"attributes":275},"6a8e6cd31807b","\n\u003Cp class=\"wp-block-paragraph\">The thing I would probably more relate the neural fusion process like is similar to the\u003Ca href=\"https:\u002F\u002Fartclasscurator.com\u002Fsurrealist-games\u002F\"> “games” surrealists used\u003C\u002Fa> to create their work from some kind of primal unconscious. One of their more famous games is called “Exquisite Corpse”. A huge sheet of paper is folded up. One person will draw on one half of the paper, and then hand it off to someone else who hasn’t seen what they’ve drawn. The fold the paper such that the other person can see a couple of the starts of lines etc but they can’t actually see the picture, and then they have to draw their own picture in such a way that it melds with all the lines on the edge.\u003C\u002Fp>\n",{},{"name":63,"clientId":277,"parentClientId":65,"renderedHtml":278,"attributes":279},"6a8e6cd31807c","\n\u003Cp class=\"wp-block-paragraph\">Then you unfold the paper and you get a weird crazy image.\u003C\u002Fp>\n",{},{"name":63,"clientId":281,"parentClientId":65,"renderedHtml":282,"attributes":283},"6a8e6cd31807d","\n\u003Cp class=\"wp-block-paragraph\">Another example is the “cut ups” that Beat Generation writers like Jack Kerouc or Williams S. Burroughs would make. They’d write a bunch of words and then they’d have to write a story using a bunch of those words re-arranged.\u003C\u002Fp>\n",{},{"name":63,"clientId":285,"parentClientId":65,"renderedHtml":286,"attributes":287},"6a8e6cd31807e","\n\u003Cp class=\"wp-block-paragraph\">Kinda like the experience of writing a doc with copy paste in an office document program, or maybe like using those magnet word tiles you see on fridges.\u003C\u002Fp>\n",{},{"name":123,"clientId":289,"parentClientId":65,"renderedHtml":290,"attributes":291},"6a8e6cd31807f","\u003Cdiv class=\"wp-block-image\">\n\u003Cfigure class=\"alignleft size-full\">\u003Cimg src=\"https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2023\u002F04\u002Fmedium-21a104cf624a5b5973f095398c1d1d98.jpg\" alt=\"\" class=\"wp-image-700\"\u002F>\u003Cfigcaption class=\"wp-element-caption\">\u003Ca href=\"https:\u002F\u002Fpleated-jeans.com\u002F2015\u002F06\u002F01\u002Fdepressing-fridge-poems-25-pics\u002F\">source\u003C\u002Fa>\u003C\u002Ffigcaption>\u003C\u002Ffigure>\n\u003C\u002Fdiv>",{},{"name":63,"clientId":293,"parentClientId":65,"renderedHtml":294,"attributes":295},"6a8e6cd318080","\n\u003Cp class=\"wp-block-paragraph\">Working with the neural network is a lot like that. You have a partner who is over the side of the wall and you can’t see them, but you can kinda hear them a little bit, and maybe they speak a different language, but you can kinda interact. Working with a neural network feels a lot more to me like improv comedy than it does using a filter.\u003C\u002Fp>\n",{},{"name":63,"clientId":297,"parentClientId":65,"renderedHtml":298,"attributes":299},"6a8e6cd318082","\n\u003Cp class=\"wp-block-paragraph\">You curate the source images and your partner transforms them in unexpected way. The next step then is to actually take those resulting images, curate them, layer them together in Photoshop, and selectively pull in and out different aspects of these paintings while also adding your own bespoke touches.\u003C\u002Fp>\n",{},{"name":86,"clientId":301,"parentClientId":65,"renderedHtml":302,"attributes":303},"6a8e6cd318085","\n\u003Ch2 class=\"wp-block-heading\">Finally painting\u003C\u002Fh2>\n",{},{"name":86,"clientId":305,"parentClientId":65,"renderedHtml":306,"attributes":307},"6a8e6cd318088","\n\u003Ch3 class=\"wp-block-heading\">Layer Structure\u003C\u002Fh3>\n",{},{"name":63,"clientId":309,"parentClientId":65,"renderedHtml":310,"attributes":311},"6a8e6cd31808b","\n\u003Cp class=\"wp-block-paragraph\">The basic layer structure is simple.\u003C\u002Fp>\n",{},{"name":63,"clientId":313,"parentClientId":65,"renderedHtml":314,"attributes":315},"6a8e6cd31808c","\n\u003Cp class=\"wp-block-paragraph\">It’s basically just the original photo with a bunch of different NN renders laid over top, with selectively revealing and hiding different parts of the renders using layer masks. On top of all that you put in some key details to “sell” the painted look and help keep things clean, enhance the art direction etc.\u003C\u002Fp>\n",{},{"name":63,"clientId":317,"parentClientId":65,"renderedHtml":318,"attributes":319},"6a8e6cd31808e","\n\u003Cp class=\"wp-block-paragraph\">To run through it with an example:\u003C\u002Fp>\n",{},{"name":63,"clientId":321,"parentClientId":65,"renderedHtml":322,"attributes":323},"6a8e6cd31808f","\n\u003Cp class=\"wp-block-paragraph\">Original source photo on the bottom.\u003C\u002Fp>\n",{},{"name":325,"clientId":326,"parentClientId":65,"renderedHtml":327,"attributes":328},"core\u002Fgallery","6a8e6cd318091","\n\u003Cfigure class=\"wp-block-gallery has-nested-images columns-default is-cropped wp-block-gallery-1 is-layout-flex wp-block-gallery-is-layout-flex\">\n\u003Cfigure class=\"wp-block-image size-large\">\u003Cimg data-id=\"701\" src=\"https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2023\u002F04\u002Fmedium-1d684863aa0128cfe30a80e70f76ae28.jpg\" alt=\"\" class=\"wp-image-701\"\u002F>\u003C\u002Ffigure>\n\n\n\n\u003Cfigure class=\"wp-block-image size-large\">\u003Cimg data-id=\"702\" src=\"https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2023\u002F04\u002Fmedium-d4f56b7e51ea90474d684a2c92ac4180.jpg\" alt=\"\" class=\"wp-image-702\"\u002F>\u003C\u002Ffigure>\n\u003Cfigcaption class=\"blocks-gallery-caption wp-element-caption\">It's very subtle but this is the denoised base layer created via filters.\u003C\u002Ffigcaption>\u003C\u002Ffigure>\n",{},{"name":123,"clientId":330,"parentClientId":326,"renderedHtml":331,"attributes":332},"6a8e6cd318094","\n\u003Cfigure class=\"wp-block-image size-large\">\u003Cimg src=\"https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2023\u002F04\u002Fmedium-1d684863aa0128cfe30a80e70f76ae28.jpg\" alt=\"\" class=\"wp-image-701\"\u002F>\u003C\u002Ffigure>\n",{},{"name":123,"clientId":334,"parentClientId":326,"renderedHtml":335,"attributes":336},"6a8e6cd318096","\n\u003Cfigure class=\"wp-block-image size-large\">\u003Cimg src=\"https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2023\u002F04\u002Fmedium-d4f56b7e51ea90474d684a2c92ac4180.jpg\" alt=\"\" class=\"wp-image-702\"\u002F>\u003C\u002Ffigure>\n",{},{"name":63,"clientId":338,"parentClientId":65,"renderedHtml":339,"attributes":340},"6a8e6cd318099","\n\u003Cp class=\"wp-block-paragraph\">Then a version of the photo that has been edited using photoshop. The primary goal of this layer is to create a denoised non-photographic layer that can act as a base. You can create this layer with a combination of filters. You can do it however you want but my approach is to start stuff out with filter>pixelate>facet and then one of the filters from the filter galleries like drybrush.\u003C\u002Fp>\n",{},{"name":63,"clientId":342,"parentClientId":65,"renderedHtml":343,"attributes":344},"6a8e6cd31809c","\n\u003Cp class=\"wp-block-paragraph\">Then I find the most important details that have been lost to this process and I use a layer mask to hide those areas so that the photo shows through. The difference is subtle but important.\u003C\u002Fp>\n",{},{"name":123,"clientId":346,"parentClientId":65,"renderedHtml":347,"attributes":348},"6a8e6cd31809e","\n\u003Cfigure class=\"wp-block-image alignwide size-full\">\u003Cimg src=\"https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2023\u002F04\u002Fmedium-7212fca7e130f1f8a5f6ad3c9c2a52f4.png\" alt=\"\" class=\"wp-image-703\"\u002F>\u003C\u002Ffigure>\n",{},{"name":63,"clientId":350,"parentClientId":65,"renderedHtml":351,"attributes":352},"6a8e6cd3180a1","\n\u003Cp class=\"wp-block-paragraph\">Then just layer up all the neural-network outputs as you like on top. I’ve isolated them each here below so you can see what they look like.\u003C\u002Fp>\n",{},{"name":325,"clientId":354,"parentClientId":65,"renderedHtml":355,"attributes":356},"6a8e6cd3180a4","\n\u003Cfigure class=\"wp-block-gallery has-nested-images columns-4 is-cropped wp-block-gallery-2 is-layout-flex wp-block-gallery-is-layout-flex\">\n\u003Cfigure class=\"wp-block-image size-large\">\u003Cimg data-id=\"704\" src=\"https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2023\u002F04\u002Fmedium-d20f882fe40a07f58132fc23b7095381.jpg\" alt=\"\" class=\"wp-image-704\"\u002F>\u003C\u002Ffigure>\n\n\n\n\u003Cfigure class=\"wp-block-image size-large\">\u003Cimg data-id=\"705\" src=\"https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2023\u002F04\u002Fmedium-ca0720de3b24bf1abdae615bfd3daf54.jpg\" alt=\"\" class=\"wp-image-705\"\u002F>\u003C\u002Ffigure>\n\n\n\n\u003Cfigure class=\"wp-block-image size-large\">\u003Cimg data-id=\"707\" src=\"https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2023\u002F04\u002Fmedium-48d07a3eb3b3310fda3997199f8abdf8.jpg\" alt=\"\" class=\"wp-image-707\"\u002F>\u003C\u002Ffigure>\n\n\n\n\u003Cfigure class=\"wp-block-image size-large\">\u003Cimg data-id=\"706\" src=\"https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2023\u002F04\u002Fmedium-a6526d36237df67ccd6bc6c94dd28e23.jpg\" alt=\"\" class=\"wp-image-706\"\u002F>\u003C\u002Ffigure>\n\n\n\n\u003Cfigure class=\"wp-block-image size-large\">\u003Cimg data-id=\"708\" src=\"https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2023\u002F04\u002Fmedium-740153d400583dffb32385b41e8f48f7.jpg\" alt=\"\" class=\"wp-image-708\"\u002F>\u003C\u002Ffigure>\n\n\n\n\u003Cfigure class=\"wp-block-image size-large\">\u003Cimg data-id=\"709\" src=\"https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2023\u002F04\u002Fmedium-7beca402ad872199c81f604a8175bfc4.jpg\" alt=\"\" class=\"wp-image-709\"\u002F>\u003C\u002Ffigure>\n\n\n\n\u003Cfigure class=\"wp-block-image size-full\">\u003Cimg data-id=\"710\" src=\"https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2023\u002F04\u002Fmedium-e4ed7b0dde1ebf0a96a30b7313a9175c.jpg\" alt=\"\" class=\"wp-image-710\"\u002F>\u003C\u002Ffigure>\n\u003C\u002Ffigure>\n",{},{"name":123,"clientId":358,"parentClientId":354,"renderedHtml":359,"attributes":360},"6a8e6cd3180a7","\n\u003Cfigure class=\"wp-block-image size-large\">\u003Cimg src=\"https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2023\u002F04\u002Fmedium-d20f882fe40a07f58132fc23b7095381.jpg\" alt=\"\" class=\"wp-image-704\"\u002F>\u003C\u002Ffigure>\n",{},{"name":123,"clientId":362,"parentClientId":354,"renderedHtml":363,"attributes":364},"6a8e6cd3180a9","\n\u003Cfigure class=\"wp-block-image size-large\">\u003Cimg src=\"https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2023\u002F04\u002Fmedium-ca0720de3b24bf1abdae615bfd3daf54.jpg\" alt=\"\" class=\"wp-image-705\"\u002F>\u003C\u002Ffigure>\n",{},{"name":123,"clientId":366,"parentClientId":354,"renderedHtml":367,"attributes":368},"6a8e6cd3180ab","\n\u003Cfigure class=\"wp-block-image size-large\">\u003Cimg src=\"https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2023\u002F04\u002Fmedium-48d07a3eb3b3310fda3997199f8abdf8.jpg\" alt=\"\" class=\"wp-image-707\"\u002F>\u003C\u002Ffigure>\n",{},{"name":123,"clientId":370,"parentClientId":354,"renderedHtml":371,"attributes":372},"6a8e6cd3180ac","\n\u003Cfigure class=\"wp-block-image size-large\">\u003Cimg src=\"https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2023\u002F04\u002Fmedium-a6526d36237df67ccd6bc6c94dd28e23.jpg\" alt=\"\" class=\"wp-image-706\"\u002F>\u003C\u002Ffigure>\n",{},{"name":123,"clientId":374,"parentClientId":354,"renderedHtml":375,"attributes":376},"6a8e6cd3180ae","\n\u003Cfigure class=\"wp-block-image size-large\">\u003Cimg src=\"https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2023\u002F04\u002Fmedium-740153d400583dffb32385b41e8f48f7.jpg\" alt=\"\" class=\"wp-image-708\"\u002F>\u003C\u002Ffigure>\n",{},{"name":123,"clientId":378,"parentClientId":354,"renderedHtml":379,"attributes":380},"6a8e6cd3180af","\n\u003Cfigure class=\"wp-block-image size-large\">\u003Cimg src=\"https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2023\u002F04\u002Fmedium-7beca402ad872199c81f604a8175bfc4.jpg\" alt=\"\" class=\"wp-image-709\"\u002F>\u003C\u002Ffigure>\n",{},{"name":123,"clientId":382,"parentClientId":354,"renderedHtml":383,"attributes":384},"6a8e6cd3180b0","\n\u003Cfigure class=\"wp-block-image size-full\">\u003Cimg src=\"https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2023\u002F04\u002Fmedium-e4ed7b0dde1ebf0a96a30b7313a9175c.jpg\" alt=\"\" class=\"wp-image-710\"\u002F>\u003C\u002Ffigure>\n",{},{"name":63,"clientId":386,"parentClientId":65,"renderedHtml":387,"attributes":388},"6a8e6cd3180b2","\n\u003Cp class=\"wp-block-paragraph\">The more you are using the neural-network output for detail the higher up it will probably be in the layer structure. The more “broad” the output render the lower down it will be. This is pretty basic though, it’s not a hard and fast rule or anything. Just use your brain and don’t have things on top that should be hidden behind other layers.\u003C\u002Fp>\n",{},{"name":63,"clientId":390,"parentClientId":65,"renderedHtml":391,"attributes":392},"6a8e6cd3180b6","\n\u003Cp class=\"wp-block-paragraph\">This isn’t a linear process either. It’s a lot of hopping around and pulling in and out various layers while also sometimes creating new layers to do custom detail work on. Here’s what all the hand painted detail looks like in isolation:\u003C\u002Fp>\n",{},{"name":123,"clientId":394,"parentClientId":65,"renderedHtml":395,"attributes":396},"6a8e6cd3180b8","\u003Cdiv class=\"wp-block-image\">\n\u003Cfigure class=\"alignleft size-full\">\u003Cimg src=\"https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2026\u002F05\u002F1-dmg5O69ajl5NbDKPySjHQ.jpeg\" alt=\"\" class=\"wp-image-615\"\u002F>\u003Cfigcaption class=\"wp-element-caption\">Adding hand-painted detail to author key areas.\u003C\u002Ffigcaption>\u003C\u002Ffigure>\n\u003C\u002Fdiv>",{},{"name":63,"clientId":398,"parentClientId":65,"renderedHtml":399,"attributes":400},"6a8e6cd3180bb","\n\u003Cp class=\"wp-block-paragraph\">Neural networks provide “specificity plus abstraction” that photoshop filters do not. This is something I discussed in\u003Ca href=\"https:\u002F\u002Fwww.jackalope.tech\u002Fhow-artists-can-use-neural-networks-to-make-art\u002F\"> my first article\u003C\u002Fa> on this subject. Adding custom detail is the last step in this neural network painting project and it’s a vital one because it allows for specificity which is beyond what even artificial neural networks provide. It uses the most powerful neural network of all! The human mind! It’s the thing that makes all the difference between “neural network as a fancy Photoshop filter” and “neural network as an assistive tool for artists”.\u003C\u002Fp>\n",{},{"name":63,"clientId":402,"parentClientId":65,"renderedHtml":403,"attributes":404},"6a8e6cd3180be","\n\u003Cp class=\"wp-block-paragraph\">In addition to providing an additional higher level of specificity that the artitificial neural network is not capable of on it’s own, at this time, adding in custom detail gives the artist another step in which to inject their own expressive subjectivity. As I said in my first article, the ability to abstract and maintain specificity is part of what makes a piece of art look “human”. This is because choices must be made for what information is eliminated and what information is emphasized. In the same way that you participate in a dialogue with the neural network to pull out and push back various “versions” of the picture through the use of masks to selectively hide and reveal, so too do you further express your own subjectivity through the choices you make for what details to manually paint in, and which you choose to remove.\u003C\u002Fp>\n",{},{"name":63,"clientId":406,"parentClientId":65,"renderedHtml":407,"attributes":408},"6a8e6cd3180bf","\n\u003Cp class=\"wp-block-paragraph\">Some examples of detailing you might want to consider using at this stage is:\u003C\u002Fp>\n",{},{"name":410,"clientId":411,"parentClientId":65,"renderedHtml":412,"attributes":413},"core\u002Flist","6a8e6cd3180c3","\n\u003Cul class=\"wp-block-list\">\n\u003Cli>pull out foreground from background to emphasize different elements, increase perceived depth, etc\u003C\u002Fli>\n\n\n\n\u003Cli>Sharpen\u002Fhighlight edges to better define the form\u003C\u002Fli>\n\n\n\n\u003Cli>smooth over weird textures that throw off the wholeness and solidity of an images form\u003C\u002Fli>\n\n\n\n\u003Cli>sharpen eyes (and other facial features) and add\u003Ca href=\"https:\u002F\u002Fascmag.com\u002Fblog\u002Fshot-craft\u002Feye-lights\"> eye-light\u003C\u002Fa>\u003C\u002Fli>\n\u003C\u002Ful>\n",{},{"name":415,"clientId":416,"parentClientId":411,"renderedHtml":417,"attributes":418},"core\u002Flist-item","6a8e6cd3180c4","\n\u003Cli>pull out foreground from background to emphasize different elements, increase perceived depth, etc\u003C\u002Fli>\n",{},{"name":415,"clientId":420,"parentClientId":411,"renderedHtml":421,"attributes":422},"6a8e6cd3180c5","\n\u003Cli>Sharpen\u002Fhighlight edges to better define the form\u003C\u002Fli>\n",{},{"name":415,"clientId":424,"parentClientId":411,"renderedHtml":425,"attributes":426},"6a8e6cd3180c7","\n\u003Cli>smooth over weird textures that throw off the wholeness and solidity of an images form\u003C\u002Fli>\n",{},{"name":415,"clientId":428,"parentClientId":411,"renderedHtml":429,"attributes":430},"6a8e6cd3180c8","\n\u003Cli>sharpen eyes (and other facial features) and add\u003Ca href=\"https:\u002F\u002Fascmag.com\u002Fblog\u002Fshot-craft\u002Feye-lights\"> eye-light\u003C\u002Fa>\u003C\u002Fli>\n",{},{"name":63,"clientId":432,"parentClientId":65,"renderedHtml":433,"attributes":434},"6a8e6cd3180cb","\n\u003Cp class=\"wp-block-paragraph\">Finally with the details added to the neural network layers together:\u003C\u002Fp>\n",{},{"name":123,"clientId":436,"parentClientId":65,"renderedHtml":437,"attributes":438},"6a8e6cd3180ce","\n\u003Cfigure class=\"wp-block-image size-full\">\u003Cimg src=\"https:\u002F\u002Fwww.aslanfrench.work\u002Fmedia\u002F2026\u002F05\u002F0yRc57WDtnNleqAiX.gif\" alt=\"\" class=\"wp-image-616\"\u002F>\u003C\u002Ffigure>\n",{},{"name":86,"clientId":440,"parentClientId":65,"renderedHtml":441,"attributes":442},"6a8e6cd3180d0","\n\u003Ch2 class=\"wp-block-heading\">Sum up and the Future\u003C\u002Fh2>\n",{},{"name":63,"clientId":444,"parentClientId":65,"renderedHtml":445,"attributes":446},"6a8e6cd3180d3","\n\u003Cp class=\"wp-block-paragraph\">In summary:\u003C\u002Fp>\n",{},{"name":410,"clientId":448,"parentClientId":65,"renderedHtml":449,"attributes":450},"6a8e6cd3180d5","\n\u003Cul class=\"wp-block-list\">\n\u003Cli>Pick good style source images\u003C\u002Fli>\n\n\n\n\u003Cli>Treat your curation process of the images like it’s a dialogue with a creative partner\u003C\u002Fli>\n\n\n\n\u003Cli>Selectively pull in and out the image results according to this dialogue process\u003C\u002Fli>\n\n\n\n\u003Cli>Lock up tight the final fit and finish of the piece with manual hand detailing that supports the subjective viewpoint that the piece has produced\u003C\u002Fli>\n\u003C\u002Ful>\n",{},{"name":415,"clientId":452,"parentClientId":448,"renderedHtml":453,"attributes":454},"6a8e6cd3180d7","\n\u003Cli>Pick good style source images\u003C\u002Fli>\n",{},{"name":415,"clientId":456,"parentClientId":448,"renderedHtml":457,"attributes":458},"6a8e6cd3180d9","\n\u003Cli>Treat your curation process of the images like it’s a dialogue with a creative partner\u003C\u002Fli>\n",{},{"name":415,"clientId":460,"parentClientId":448,"renderedHtml":461,"attributes":462},"6a8e6cd3180db","\n\u003Cli>Selectively pull in and out the image results according to this dialogue process\u003C\u002Fli>\n",{},{"name":415,"clientId":464,"parentClientId":448,"renderedHtml":465,"attributes":466},"6a8e6cd3180dc","\n\u003Cli>Lock up tight the final fit and finish of the piece with manual hand detailing that supports the subjective viewpoint that the piece has produced\u003C\u002Fli>\n",{},{"name":63,"clientId":468,"parentClientId":65,"renderedHtml":469,"attributes":470},"6a8e6cd3180de","\n\u003Cp class=\"wp-block-paragraph\">Whew this has been a big series for me. I’ve got a lot more articles in progress on the subject of neural networks, technology, art, and the intersection of art and technology. I am going to change my approach though away from these multipart series into something which is more agile and which I can get updated more often.\u003C\u002Fp>\n",{},{"name":63,"clientId":472,"parentClientId":65,"renderedHtml":473,"attributes":474},"6a8e6cd3180e1","\n\u003Cp class=\"wp-block-paragraph\">I’m learning a bunch of stuff super fast, and trying to cram everything into a single monolithic manifesto is too slow a way to work. I do have a lot of stuff related to this that I want to talk about though, so if you’re interested in keeping up on the subject,\u003Ca href=\"https:\u002F\u002Fmailchi.mp\u002Ff75162f1141a\u002Fdesign-x-code_newsletter\"> please subscribe to the newsletter\u003C\u002Fa>! (2023 update: you can still subscribe! I never post though!)\u003C\u002Fp>\n",{},{"name":63,"clientId":476,"parentClientId":65,"renderedHtml":477,"attributes":478},"6a8e6cd3180e3","\n\u003Cp class=\"wp-block-paragraph\">As a teaser of what comes next, I’m thinking of trying to do some proof of concept work using neural networks to “photobash”. Photobashing is a digital painting technique often used in the concept art world to generate imagery extremely fast. I think there’s a lot of possibility here for using neural networks as a part of this process. So expect some more fantastical pieces soon that aren’t just straight ahead portraits!\u003C\u002Fp>\n",{},{"name":480,"clientId":481,"parentClientId":65,"renderedHtml":482,"attributes":483},"core\u002Fembed","6a8e6cd3180e4","\n\u003Cfigure class=\"wp-block-embed is-type-video is-provider-vimeo wp-block-embed-vimeo\">\u003Cdiv class=\"wp-block-embed__wrapper\">\nhttps:\u002F\u002Fvimeo.com\u002F5666254\n\u003C\u002Fdiv>\u003C\u002Ffigure>\n",{},{"name":485,"clientId":486,"parentClientId":65,"renderedHtml":487,"attributes":488},"core\u002Ffootnotes","6a8e6cd3180e7","\u003Col class=\"wp-block-footnotes\">\u003Cli id=\"bfb2f344-67dd-47a2-b017-8de6ddc9f325\">Actually not totally true as apparently your rods can detect some blue light at extremely low light conditions. Neato! \u003Ca href=\"#bfb2f344-67dd-47a2-b017-8de6ddc9f325-link\" aria-label=\"Jump to footnote reference 1\">↩︎\u003C\u002Fa>\u003C\u002Fli>\u003Cli id=\"5fd6cd63-cc21-4290-9955-fa3ca565485b\">I have an essay I want to write about this more specifically but I’ll forgo it for now.\u003Cbr> \u003Ca href=\"#5fd6cd63-cc21-4290-9955-fa3ca565485b-link\" aria-label=\"Jump to footnote reference 2\">↩︎\u003C\u002Fa>\u003C\u002Fli>\u003Cli id=\"3ba9ec09-18fc-4ef8-9c94-24b27b8ae85b\">This is a bit simplistic but I’ll cover this issue in a later blog specifically about generative art and it’s relationship to neural networks. \u003Ca href=\"#3ba9ec09-18fc-4ef8-9c94-24b27b8ae85b-link\" aria-label=\"Jump to footnote reference 3\">↩︎\u003C\u002Fa>\u003C\u002Fli>\u003C\u002Fol>",{},{"name":63,"clientId":490,"parentClientId":65,"renderedHtml":491,"attributes":492},"6a8e6cd3180ea","\n\u003Cp class=\"wp-block-paragraph\">\u003C\u002Fp>\n",{},1787718861263]