{"id":1754,"date":"2023-11-27T18:12:43","date_gmt":"2023-11-27T18:12:43","guid":{"rendered":"https:\/\/denayer.com\/blog\/?p=1754"},"modified":"2024-02-03T10:21:37","modified_gmt":"2024-02-03T10:21:37","slug":"guilty-pleasure","status":"publish","type":"post","link":"http:\/\/denayer.com\/blog\/guilty-pleasure\/","title":{"rendered":"Guilty Pleasure &#8211; comparing DALL-E, MidJourney, Stable Diffusion and Photoshop"},"content":{"rendered":"<p>[et_pb_section fb_built=&#8221;1&#8243; _builder_version=&#8221;4.23.1&#8243; _module_preset=&#8221;default&#8221; global_colors_info=&#8221;{}&#8221;][et_pb_row _builder_version=&#8221;4.23.1&#8243; _module_preset=&#8221;default&#8221; global_colors_info=&#8221;{}&#8221;][et_pb_column type=&#8221;4_4&#8243; _builder_version=&#8221;4.23.1&#8243; _module_preset=&#8221;default&#8221; global_colors_info=&#8221;{}&#8221;][et_pb_text _builder_version=&#8221;4.23.1&#8243; _module_preset=&#8221;default&#8221; hover_enabled=&#8221;0&#8243; global_colors_info=&#8221;{}&#8221; sticky_enabled=&#8221;0&#8243;]<\/p>\n<p>AI-based image generation tools are transforming the way we perceive and create art. I explored four such tools to understand their unique capabilities and limitations. The test subject was the <a href=\"https:\/\/www.dame-blanche.be\" target=\"_blank\" rel=\"noopener\">dame blanche website<\/a> (and <a href=\"https:\/\/www.instagram.com\/dameblancheart\/\" target=\"_blank\" rel=\"noopener\">Instagram feed<\/a>) that I have finally released after boring my friends with the silly idea for years. <span>Join the discussion on <a href=\"https:\/\/www.linkedin.com\/feed\/update\/urn:li:activity:7134970270884220932\/\" target=\"_blank\" rel=\"noopener\">LinkedIn<\/a>.<\/span><\/p>\n<p>That said image generation has big potential for business in marketing and advertising so time to explore the tools of the trade!<\/p>\n<p>in summary:<\/p>\n<ul>\n<li>DALL-E is super easy to use with high quality images that don\u2019t look realistic<\/li>\n<li>MidJourney gives more creative control and can create realistic images<\/li>\n<li>Stable Diffusion is for hackers with full control but less quality<\/li>\n<li>Photoshop created a super nice image but ignored the finer elements of the description<\/li>\n<\/ul>\n<p>for the comparison I used this prompt: \u201cdame blanche ice cream with dripping chocolate sauce in style of Renaissance\u201d. I then used this base prompt to ask ChatGPT to generate a much more detailed description (pro tip). I used this expanded prompt for all trials, except for DALL-E where this is automatically done under the hood.<\/p>\n<p>[\/et_pb_text][et_pb_image src=&#8221;http:\/\/denayer.com\/blog\/wp-content\/uploads\/2023\/11\/db_comparison_final.jpg&#8221; _builder_version=&#8221;4.23.1&#8243; _module_preset=&#8221;default&#8221; global_colors_info=&#8221;{}&#8221;][\/et_pb_image][et_pb_text _builder_version=&#8221;4.23.1&#8243; _module_preset=&#8221;default&#8221; global_colors_info=&#8221;{}&#8221;]<\/p>\n<p>DALL-E 3<\/p>\n<ul>\n<li>easy to use, integrated into OpenAI ChatGPT Plus, has a really good understanding of the request, with minimal detail, just ask for an image and a visually appealing image comes out. ChatGPT will first rewrite your prompt before sending it to DALL-E, which is nice.\n<ul>\n<li>A sumptuous depiction of a Dame Blanche ice cream, characterized by its creamy texture and elegant presentation. The ice cream is topped with rich\u2026<\/li>\n<\/ul>\n<\/li>\n<li>limited creative control leading to many images looking the same<\/li>\n<li>no way to consistently generate images in the same style (unless you stay in the same chat session). This is a big limitation when using image generation for professional use where you may want a common style in a series of images.<\/li>\n<li>The DALL-E editor (only for v2) lets you manipulate generated images, eg by removing a section and asking DALL-E to in paint the missing part.<\/li>\n<li>the OpenAI content policy blocks many creative avenues, even if they are perfectly legal, for example when copyright of an artist has already expired.<\/li>\n<li>automation is possible through the DALL-E API, however, for some reason the same prompts sent through the API consistently lead to lower quality images when comparing with the web interface<\/li>\n<\/ul>\n<p>MidJourney 5.2<\/p>\n<ul>\n<li>a bit harder to use, but with more creative control, you can specifiy lighting, perspective, and camera type. You can refer to the typical style of a director, a film genre, level of creativity, etc. A prompt would look something like this: A Cinematic scene from [YEAR, MOVIE GENRE, MOVIE NAME], [SHOT TYPE], [SCENE\/SUBJECT\/ACTION] captured by [CINEMATIC CAMERA], film directed by [DIRECTOR], [EMOTIONI, [LIGHTING] &#8211;ar 1:1 &#8211;style raw \u2014-v 5.1<\/li>\n<li>can refer to the style of living artists (which is interesting because copying a style is ok but the images shouldn\u2019t be in the training data so it should be impossible\u2026)<\/li>\n<li>you can upload your own image and have them modified<\/li>\n<li>you can consistently generate images by controlling the \u2018seed\u2019. The seed determines the starting point for the randomisation of the neural network. &#8211; controlling this starting point gives you the predictability to create images in the same style.<\/li>\n<li>you can see live what others are creating and how they do it (with a higher subscription you can also stay hidden if you want)<\/li>\n<\/ul>\n<p>Stable Diffusion SDXL 1<\/p>\n<ul>\n<li>open source, can run locally (I run it on an RTX3090) and SaaS<\/li>\n<li>many people share their custom fine tuned models and what they have created, along with the prompt and the settings so you can learn from them<\/li>\n<li>full control over the models and all parameters of the image generation<\/li>\n<li>consistently generate images through control of the seed (see above)<\/li>\n<li>can automate with the API at <a href=\"http:\/\/stability.ai\/\">stability.ai<\/a><\/li>\n<\/ul>\n<p>Adobe Photoshop 2024<\/p>\n<ul>\n<li>super easy to use and seamlessly integrated into the Photoshop user interface. Uses the Photoshop image library as training set.<\/li>\n<li>can replace \/ remove parts of an image (in-painting \/ generative fill)<\/li>\n<li>excellent for extending an image to make it bigger by adding missing environment (outpainting)<\/li>\n<li>seems still early days in terms of creative control + there is a 500 character prompt length limitation so you cannot fully use the extended description generated by chatgpt<\/li>\n<\/ul>\n<p>what are your experiences with these tools? Comment below!<\/p>\n<p>[\/et_pb_text][\/et_pb_column][\/et_pb_row][\/et_pb_section]<\/p>\n","protected":false},"excerpt":{"rendered":"<p>AI-based image generation tools are transforming the way we perceive and create art. I explored four such tools to understand their unique capabilities and limitations. The test subject was the dame blanche website (and Instagram feed) that I have finally released after boring my friends with the silly idea for years. Join the discussion on [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":1762,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_et_pb_use_builder":"on","_et_pb_old_content":"","_et_gb_content_width":"","footnotes":""},"categories":[34],"tags":[],"class_list":["post-1754","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai"],"aioseo_notices":[],"_links":{"self":[{"href":"http:\/\/denayer.com\/blog\/wp-json\/wp\/v2\/posts\/1754","targetHints":{"allow":["GET"]}}],"collection":[{"href":"http:\/\/denayer.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/denayer.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/denayer.com\/blog\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"http:\/\/denayer.com\/blog\/wp-json\/wp\/v2\/comments?post=1754"}],"version-history":[{"count":0,"href":"http:\/\/denayer.com\/blog\/wp-json\/wp\/v2\/posts\/1754\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"http:\/\/denayer.com\/blog\/wp-json\/wp\/v2\/media\/1762"}],"wp:attachment":[{"href":"http:\/\/denayer.com\/blog\/wp-json\/wp\/v2\/media?parent=1754"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/denayer.com\/blog\/wp-json\/wp\/v2\/categories?post=1754"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/denayer.com\/blog\/wp-json\/wp\/v2\/tags?post=1754"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}