{"id":4924,"date":"2023-04-23T14:52:59","date_gmt":"2023-04-23T21:52:59","guid":{"rendered":"http:\/\/www.contrapositivediary.com\/?p=4924"},"modified":"2023-04-23T16:40:52","modified_gmt":"2023-04-23T23:40:52","slug":"ai-image-generators-mon-dieu","status":"publish","type":"post","link":"https:\/\/www.contrapositivediary.com\/?p=4924","title":{"rendered":"AI Image Generators, Mon Dieu"},"content":{"rendered":"

I finished a 10,700 novelette the other day, the first short fiction I\u2019ve finished since 2008, when I wrote \u201cSympathy on the Loss of One of Your Legs,\u201d now available in my collection, Souls in Silicon<\/a><\/em>. I\u2019ve mostly written novels and short novels since then. (I’ll have more to say about \u201cVolare\u201d in a future entry here.)<\/p>\n

To be published, it needs a cover. I have no objection to paying artists for covers, which apart from an experiment or two (see \u201cWhale Meat\u201d) I\u2019ve always done in the past. Given all the yabbjabber about AI content creation recently, I thought, \u201cHey, here\u2019s a chance to see if it\u2019s all BS.\u201d<\/p>\n

The spoiler: It\u2019s not all BS, but parts of it are BS-ier than others.<\/p>\n

Ok. I\u2019ve tested two AI image generators: OpenAI\u2019s DALL-E 2, and Microsft\u2019s Bing Image Generator. I found them through a solid article on ZDNet by Sabrina Ortiz<\/a>. As it happens, Bing Image Generator outsources the process to DALL-E. I wanted to try Midjourney, and may eventually, but you have to have a paid subscription (about $8\/month) to use it.<\/p>\n

I\u2019m not going to summarize the story here. One image I wanted to try as a cover would be the female lead sitting with her behind in a wicker basket, floating through the air at dawn a thousand feet or so over Baltimore. In both generators (which are basically the same generator) you feed the AI a detailed text description and turn it loose. I started simple: \u201cA woman flying through the air in a wicker basket.\u201d Edy Gagliano does precisely that in the story. What DALL-E gave me was this:<\/p>\n

\"DALL\u00b7E<\/a><\/p>\n

Well, the woman is flying through the air, but we have a preposition problem here. She is over<\/em>, not in<\/em> the basket. Good first shot, though. I tried various extensions of that basic description, to the tune of 48 images on Dall-E. I won\u2019t post them all here for space reasons, but they ran the gamut: A woman flying through the air holding a basket, a woman flying through the air in a basket the size and shape of a bathtub, and on and on.<\/p>\n

The next one here is perhaps the best I\u2019ve gotten from DALL-E. It\u2019s a woman in a basket over Baltimore, I guess. Here\u2019s the description: \u201ca barefoot woman sitting down inside a magical wicker basket that flies through the air at dawn over Baltimore.\u201d In one sense, it\u2019s not a bad picture:<\/p>\n

\"DALL\u00b7E<\/a><\/p>\n

That said, It looks out of focus. The basket is not wicker and it\u2019s yuge<\/em>. And in the story, Edy just puts her butt in the basket and lets her legs hang over the side.<\/p>\n

Now let us move over to Bing Image Generator. In a way, it came closer than nearly all of the DALL-E images. But now we confront a well-known weakness of AI image generators: They can\u2019t draw realistic hands or feet or faces. Here\u2019s my first take on the image from Bing:<\/p>\n

\"_77229ce5-3d7c-4c09-964f-b2b784ba3580<\/a><\/p>\n

Look closely. Her hands and feet appear to be drawn by something that doesn\u2019t know what a human hand or foot looks like. The face, furthermore, looks like it has one eye missing. (That\u2019s easier to see in the full-sized image.)<\/p>\n

I\u2019ll give Bing credit: The images are less fuzzy and smeary. Because Bing uses DALL-E, I suspect there are DALL-E settings I don\u2019t know about yet. I tried a few more times and got some reasonable images, all of them including some weirdness or another. The one below is a better rendering of a woman who is actually sitting in the basket with her legs hanging over the basket\u2019s edge. But did I order a helicopter? Her face is a little lopsided, and her hands and feet, while not grotesque, aren\u2019t quite<\/em> right.<\/p>\n

\"_090cd681-df9a-4736-8fcd-cdaafe028ae1<\/a><\/p>\n

Bing gave me about 24 images while I messed with it, and some of the images, while not capturing what I intended, were well-rendered and not full of weirdness. The one below is probably closest to Edy as I imagine her, and we get a SpaceX booster burning up in the atmosphere to boot. Is she over Baltimore? I don\u2019t know Baltimore well enough to be sure, but that, at least, doesn\u2019t matter. Stock photos of anonymous cities are everywhere.<\/p>\n

\"_794c2ce1-7cd6-492d-9712-7e75ab646a3c<\/a><\/p>\n

None of the others are notable enough to show here.<\/p>\n

So where does this leave us? AIs can draw pictures. That\u2019s real, and I\u2019m guessing that if you tell it to draw something a little less loopy than a woman with her butt in a flying basket, it might do a better job. I remain puzzled why hands and feet and faces are so hard to do. Don\u2019t AIs need training? And aren\u2019t there plenty of photos of hands and feet and faces for them to generalize from a substantial number of specific examples?<\/p>\n

I have no idea how these things are supposed to work, and if there were a good overview book on AI image generator internals, I\u2019d buy it like a shot. In the meantime, I may practice some more and look at specific settings. If nothing else, I can produce some concept images to show to a cover artist. And maybe I\u2019ll luck into something usable as-is.<\/p>\n

Whatever I discover, you can count on seeing it here.<\/p>\n","protected":false},"excerpt":{"rendered":"

I finished a 10,700 novelette the other day, the first short fiction I\u2019ve finished since 2008, when I wrote \u201cSympathy on the Loss of One of Your Legs,\u201d now available in my collection, Souls in Silicon. I\u2019ve mostly written novels and short novels since then. (I’ll have more to say about \u201cVolare\u201d in a future […]<\/p>\n","protected":false},"author":2,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[27],"tags":[118,255,30,256],"class_list":["post-4924","post","type-post","status-publish","format-standard","hentry","category-reviews","tag-ai","tag-bing","tag-images","tag-openai"],"_links":{"self":[{"href":"https:\/\/www.contrapositivediary.com\/index.php?rest_route=\/wp\/v2\/posts\/4924","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.contrapositivediary.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.contrapositivediary.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.contrapositivediary.com\/index.php?rest_route=\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.contrapositivediary.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=4924"}],"version-history":[{"count":18,"href":"https:\/\/www.contrapositivediary.com\/index.php?rest_route=\/wp\/v2\/posts\/4924\/revisions"}],"predecessor-version":[{"id":4954,"href":"https:\/\/www.contrapositivediary.com\/index.php?rest_route=\/wp\/v2\/posts\/4924\/revisions\/4954"}],"wp:attachment":[{"href":"https:\/\/www.contrapositivediary.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=4924"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.contrapositivediary.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=4924"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.contrapositivediary.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=4924"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}