{"id":19420,"date":"2024-09-05T11:24:29","date_gmt":"2024-09-05T03:24:29","guid":{"rendered":"https:\/\/www.1ai.net\/?p=19420"},"modified":"2024-09-05T11:24:29","modified_gmt":"2024-09-05T03:24:29","slug":"%e8%bf%98%e6%9c%89%e4%ba%ba%e4%b8%8d%e7%9f%a5%e9%81%93%e5%8f%af%e4%bb%a5%e5%9b%be%e7%89%87%e5%8f%8d%e6%8e%a8%e6%8f%90%e7%a4%ba%e8%af%8d%e5%90%97%ef%bc%8c%e4%b8%8d%e8%8a%b1%e9%92%b1%e5%85%8d%e8%b4%b9","status":"publish","type":"post","link":"https:\/\/www.1ai.net\/en\/19420.html","title":{"rendered":"Does anyone know that you can reverse the prompt words from the image? Free and easy to use image reverse model Joy Caption"},"content":{"rendered":"<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-19422\" title=\"ce113014j00sjbl0v00izd000u000kim\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2024\/09\/ce113014j00sjbl0v00izd000u000kim.jpg\" alt=\"ce113014j00sjbl0v00izd000u000kim\" width=\"1080\" height=\"738\" \/><\/p>\n<h2>Joy Caption Model Introduction<\/h2>\n<p>Today&#039;s article topic will introduce you to an excellent<a href=\"https:\/\/www.1ai.net\/en\/tag\/%e5%9b%be%e5%83%8f%e5%8f%8d%e6%8e%a8%e6%a8%a1%e5%9e%8b\" title=\"[Sees articles with labels]\" target=\"_blank\" >Image Inversion Model<\/a>:<a href=\"https:\/\/www.1ai.net\/en\/tag\/joy-caption\" title=\"_Other Organiser\" target=\"_blank\" >Joy Caption<\/a>This is by the author<strong>Fancy Feast<\/strong>The Joy Caption model developed is based on Google&#039;s SigLIP model and Meta&#039;s latest Llama3.1 model, using the Adapter adaptation mode and the well-trained excellent image reverse description LLM model. It can output corresponding image description prompts with rich details according to the user&#039;s set parameters.<\/p>\n<ul>\n<li>Google&#039;s SigLIP (Sigmoid Loss for Language Image Pre-Training) is an improved multimodal model, similar to CLIP, but with a better loss function. Download address: https:\/\/huggingface.co\/google\/siglip-so400m-patch14-384<\/li>\n<li>Meta-Llama-3.1-8B-bnb-4bit is an optimized LLM large language model based on Meta&#039;s Llama 3.1 architecture. It uses the BitsAndBytes library for 4-bit quantization, which greatly reduces memory usage while maintaining model performance and accuracy. Download address: https:\/\/huggingface.co\/unsloth\/Meta-Llama-3.1-8B-bnb-4bit.<\/li>\n<li>Online experience address: https:\/\/huggingface.co\/spaces\/fancyfeast\/joy-caption-pre-alpha<\/li>\n<\/ul>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-19423\" title=\"7e3453acj00sjbl0t002bd000u000mrm\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2024\/09\/7e3453acj00sjbl0t002bd000u000mrm.jpg\" alt=\"7e3453acj00sjbl0t002bd000u000mrm\" width=\"1080\" height=\"819\" \/><\/p>\n<p><strong>Flux Joy Caption prompts reverse push experience<\/strong><\/p>\n<p>The current community already has ComfyUI plug-in support: Comfyui_CXH_joy_caption. You only need to download the corresponding model and install the plug-in to start the experience. Due to the complex environment operation or local resource bottlenecks, the operation of large models has become a threshold for local deployment. This article will introduce a free way to use the previous article.<strong>BizyAir Plugin<\/strong>, you can easily use joy_caption&#039;s excellent image inversion capabilities without any barriers.<strong>BizyAir Plugin<\/strong>Please refer to the previous article for installation: Flux: Are you still worried about Flux consuming video memory? BizyAir does not require video memory. Local ComfyUI experience cloud resource acceleration, free Dev ultra-fast drawing experience<\/p>\n<p>If you are confident that you can handle the deployment environment, you can also try the Comfyui_CXH_joy_caption plug-in in case the free period ends. For detailed instructions, please refer to the Github plug-in homepage. Plug-in address: https:\/\/github.com\/StartHua\/Comfyui_CXH_joy_caption<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-19421\" title=\"5ce805fbj00sjbl0s000ed000u0004im\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2024\/09\/5ce805fbj00sjbl0s000ed000u0004im.jpg\" alt=\"5ce805fbj00sjbl0s000ed000u0004im\" width=\"1080\" height=\"162\" \/><\/p>\n<p>Flux Wensheng Diagram Workflow<\/p>\n<p>The Flux Wensheng graph model has been introduced in the previous article. For basic installation, please refer to the article:<a href=\"https:\/\/www.1ai.net\/en\/17231.html\/\">FLUX [Sequel]: 12B parameter 23G largest open source Wensheng graph model, Dev version directly outputs stunning pictures to appreciate<\/a><\/p>\n<ul>\n<li>Flux text | Picture + LORA + CN + prompt reverse one-click switching workflow: https:\/\/www.liblib.art\/modelinfo\/782aacd70f604da39e83368c696a02a8<\/li>\n<li>Low Video Memory - Flux-Dev - GGUF Workflow (text | image): https:\/\/www.liblib.art\/modelinfo\/bf3320e00f1649a69f5835101ef04276<\/li>\n<\/ul>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-19424\" title=\"61b17a62j00sjbl0t001ld000u000h8m\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2024\/09\/61b17a62j00sjbl0t001ld000u000h8m.jpg\" alt=\"61b17a62j00sjbl0t001ld000u000h8m\" width=\"1080\" height=\"620\" \/><\/p>\n<p>Joy Caption+Flux Wenshengtu Workflow<\/p>\n<p>Just add a BizyAir inverse node to the above Wenshengtu workflow. The workflow has been uploaded to the LIBLIB platform, the workflow address is: https:\/\/www.liblib.art\/modelinfo\/3e9699c20bea4582aada692c4adccf20<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-19425\" title=\"25fe8eadj00sjbl0s001dd000u000d1m\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2024\/09\/25fe8eadj00sjbl0s001dd000u000d1m.jpg\" alt=\"25fe8eadj00sjbl0s001dd000u000d1m\" width=\"1080\" height=\"469\" \/><\/p>\n<p><strong>Notice<\/strong>: If the image is too large to be supported, you can set it in the workflow<strong>Image scaling 0.5<\/strong>set up.<\/p>\n<p>01. Leopard print<\/p>\n<p>at the time of the death of the President of the United Nations, the Secretary-General would be responsible for the death of the President of the United Nations and for the death of the Secretary-General<\/p>\n<p>Chinese Girl, This is a high-resolution photo of an East Asian woman with long dark brown hair that falls down her back. She has a slim, curvaceous figure with a modest bust. Her complexion is smooth and porcelain. She wears a tight, long-sleeved bodysuit with a bold orange tiger print on a black background that accentuates her figure. The bodysuit fits snugly to her body, accentuating her curves. Her expression is calm and inviting, with a subtle smile and closed eyes, giving her an impression of tranquility. Her makeup is natural and understated, with the focus on highlighting her features without looking too exaggerated. The background has a soft gradient texture of beige and light brown fabrics, creating a warm and cozy atmosphere. A large glowing orb (probably a softbox light) is positioned to the side, casting a warm golden light that complements the colors of the bodysuit and background. The overall mood of this photo is intimate and serene, with the focus on the subject&#039;s calm demeanor and striking appearance. The light is soft and even, and the warm tones enhance the cozy atmosphere. The style of this photo is modern, with an emphasis on natural light and subtle and elegant poses. The woman&#039;s pose is relaxed, with her hands on her thighs, adding to the sense of calm. This image was likely taken in a studio, with great attention to light and composition. The overall aesthetic is sophisticated and visually appealing. The tiger bodysuit adds a playful, whimsical touch to the otherwise serene atmosphere. This image is a fusion of fashion and portraiture, with a focus on the subject&#039;s beauty and the creative use of lighting. The style is reminiscent of high fashion photography. The model&#039;s hands are resting on her thighs, with her fingers spread, adding a subtle, playful touch to her otherwise serene pose. This image is a wonderful fusion of fashion and portraiture, beautiful and alluring. The overall atmosphere is intimate and serene, with a focus on the subject&#039;s calm demeanor and striking appearance. The lighting is soft and even, with warm tones enhancing the cozy atmosphere. The image is modern in style, with a focus on natural light and a subtle, elegant pose. The woman&#039;s pose is relaxed, with her hands on her thighs, adding to the sense of calm. This image is a fusion of fashion and<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-19425\" title=\"25fe8eadj00sjbl0s001dd000u000d1m\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2024\/09\/25fe8eadj00sjbl0s001dd000u000d1m.jpg\" alt=\"25fe8eadj00sjbl0s001dd000u000d1m\" width=\"1080\" height=\"469\" \/><\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-19426\" title=\"de78ebdaj00sjbl0t0019d000u000kim\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2024\/09\/de78ebdaj00sjbl0t0019d000u000kim.jpg\" alt=\"de78ebdaj00sjbl0t0019d000u000kim\" width=\"1080\" height=\"738\" \/><\/p>\n<p>02. Sea Lion<\/p>\n<p>I'll be able to do the same, and I'll be able to do the same, to do the same<\/p>\n<p>In the video, two animated characters are in romantic scenes, transitioning from one scene to another. In the first frame, the male character is wearing a white shirt, dark pants, and sneakers, while the female character is wearing a red top, black skirt, and high heels. They stand together, smiling, as if in an intimate moment.<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-19427\" title=\"34833cefj00sjbl0t0018d000u000b5m\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2024\/09\/34833cefj00sjbl0t0018d000u000b5m.jpg\" alt=\"34833cefj00sjbl0t0018d000u000b5m\" width=\"1080\" height=\"401\" \/><\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-19429\" title=\"6b1abc59j00sjbl0t001od000ow00w0m\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2024\/09\/6b1abc59j00sjbl0t001od000ow00w0m.jpg\" alt=\"6b1abc59j00sjbl0t001od000ow00w0m\" width=\"896\" height=\"1152\" \/><\/p>\n<p>03. Street Performing Cat<\/p>\n<p>At the same time, the United States of America and the United States of America would be the same as the United States of America<\/p>\n<p>This is a detailed, realistic digital illustration of a cat playing guitar on a rainy street. With orange and white fur and dressed in a worn green hoodie and dark blue pants, the cat gives off a casual street performance vibe. The cat&#039;s large, round eyes and ears are erected as if listening to music. The orange guitar is carefully held in the cat&#039;s paws, with the strings and fretboard clearly visible. In the foreground, a shallow metal bowl filled with coins sits on the wet sidewalk, glistening with raindrops. The background is blurred, showing several pedestrians walking by, their faces obscured by the rain and the distance. The rain is depicted as a gentle, steady drizzle, with water droplets visible on the cat&#039;s fur and the sidewalk. The overall vibe is one of melancholy urban charm, with the cat&#039;s music contrasting with the rainy, gray surroundings. The illustration skillfully captures the textures of the cat&#039;s fur, the guitar&#039;s wood, and the wet pavement, immersing the viewer in a vivid, atmospheric scene. The colors are muted, with earthy tones and the bright orange of the guitar standing out against the drab background. The style is reminiscent of photorealistic digital art, with an attention to detail textures and lighting. The overall effect is both warm and melancholic. | The image is rich in texture and detail, with the rain adding a dynamic, interactive element to the scene. | The style is highly photorealistic, with a focus on capturing the emotional depth of the scene. | The cat\u2019s expression is calm, focused, and creative, adding a sense of pathos to the scene. | The rain adds a sense of movement and energy to the scene, highlighting the cat\u2019s performance. | The background details are subtle, with the blurred silhouettes of pedestrians adding depth to the scene. | The overall mood is contemplative and peaceful, with the cat\u2019s music providing a nice contrast to the rainy surroundings. | The illustrations skillfully capture the textures of the cat\u2019s fur, the wood of the guitar, and the wet pavement, immersing the viewer in a vivid, atmospheric scene. | The colors are muted, with earthy tones and the bright orange of the guitar standing out against the drab background. | The style is reminiscent of photorealism<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-19428\" title=\"e0d38a6fj00sjbl0s001cd000u000d2m\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2024\/09\/e0d38a6fj00sjbl0s001cd000u000d2m.jpg\" alt=\"e0d38a6fj00sjbl0s001cd000u000d2m\" width=\"1080\" height=\"470\" \/><\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-19430\" title=\"013fca06j00sjbl0t002kd000ow00w0m\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2024\/09\/013fca06j00sjbl0t002kd000ow00w0m.jpg\" alt=\"013fca06j00sjbl0t002kd000ow00w0m\" width=\"896\" height=\"1152\" \/><\/p>\n<p>04. Carrying the burden<\/p>\n<p>For the first time in the history of the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, and the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, and the United States of America, the United States of America.<\/p>\n<p>This is a fantastical digital artwork depicting a surreal scene. A huge elephant with grey skin and wrinkled textures occupies the foreground, walking across a sunlit savannah. The elephant&#039;s body is adorned with lush greenery, including a large acacia tree perched on its back, with branches spreading out to either side. The tree&#039;s leaves and branches are intricately detailed, finely textured, and feature varying shades of green. In the background, a majestic medieval-style castle rises from the elephant&#039;s back, its stone walls and towers blending into the elephant&#039;s skin. The castle&#039;s architectural style is a blend of Gothic and Romanesque styles, with pointed arches, towers, and a central keep. The castle&#039;s windows and doors are adorned with intricate stone carvings. The sky above is a warm gradient blue, with soft, fluffy clouds that seem to radiate a golden glow, reminiscent of afternoon or early morning sunlight. The overall atmosphere is one of whimsical wonder, blending fantasy and realism in a dreamlike atmosphere. The image combines detailed textures with a sense of magic and adventure. The elephants&#039; path weaves through tall grass and scattered wildflowers, adding to the peaceful, idyllic atmosphere. The style of the artwork is reminiscent of high-end digital art, with an attention to realism and intricate detail.<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-19431\" title=\"e43e7a3cj00sjbl0t001ed000u000dim\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2024\/09\/e43e7a3cj00sjbl0t001ed000u000dim.jpg\" alt=\"e43e7a3cj00sjbl0t001ed000u000dim\" width=\"1080\" height=\"486\" \/><\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-19432\" title=\"0e03902cj00sjbl0t002zd000ow00w0m\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2024\/09\/0e03902cj00sjbl0t002zd000ow00w0m.jpg\" alt=\"0e03902cj00sjbl0t002zd000ow00w0m\" width=\"896\" height=\"1152\" \/><\/p>","protected":false},"excerpt":{"rendered":"<p>Introduction to Joy Caption Model Today's article will introduce you to an excellent image backpropagation model: Joy Caption, developed by author Fancy Feast, is based on Google's SigLIP model and Meta's newest Llama3.1 model, using the Adapter Adaptation Mode. The Joy Caption model is based on Google's SigLIP model and Meta's latest Llama3.1 model, using the Adapter Adaption model, and the excellent image backpropagation describing the LLM model through careful training. According to the parameters set by the user, it can output the corresponding image description prompts with rich details. Google's SigLIP (Sigmoid Loss for Language Image Pre-Training) is an improvement of the<\/p>","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[149,144],"tags":[3853,4276,4275,837],"collection":[],"class_list":["post-19420","post","type-post","status-publish","format-standard","hentry","category-jiaocheng","category-baike","tag-flux","tag-joy-caption","tag-4275","tag-837"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/19420","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/comments?post=19420"}],"version-history":[{"count":0,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/19420\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/media?parent=19420"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/categories?post=19420"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/tags?post=19420"},{"taxonomy":"collection","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/collection?post=19420"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}