Minimax H3 Photorealistic Image Generator (LORA + Workflow)Consider to download the workflow (from 'optional downloads') for easier image extraction.A conceptual lora and workflow to create high resolution photorealistic images with Minimax H3.Remember image generation on WAN? Now you can generate images with H3 too.<hr />Photorealistic datasets and special training has made it possible to create a photorealistic image generation lora for Minimax-H3, with output up to 3840x2176.With two types of available workflows, the photorealism image lora is used to either capture the first and last frame out of a 22 frame video (Default) or generates 1 frame (Cicaloo Node+VAE).Pick either the 3780 or 3990 step variant. Lora 3990 seems to be more vibrant / contrasty and sharper detailed.<hr />Text 2 ImageThis lora has been originally developed with the intent to generate photorealistic images from text.<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/771c8b6b-eceb-4219-8b3a-e97c3be23564/original=true/771c8b6b-eceb-4219-8b3a-e97c3be23564.jpeg" />Example from Text to Image<hr />Image 2 VideoYou can also use the "Image2Video" mode to inject an input image into a new output. Results may vary, lora strength may play a factor.<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/0d2a90ea-b6fc-4181-897f-7d349929d975/original=true/0d2a90ea-b6fc-4181-897f-7d349929d975.jpeg" />All images above were made using the "Default workflow".<hr />Reference Model: With support for reference model in the default workflow (not the 1 Frame workflow), you can also use the ph0t0r34l lora with a references.<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/83fb2b6a-6d37-45e3-b89c-51c9f6cba0f2/original=true/83fb2b6a-6d37-45e3-b89c-51c9f6cba0f2.jpeg" />You can try either the FL2VA model or REF2FLVA model, both may yield different results.<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/1d7feb55-d545-4834-8ad5-d54cafb31dbd/original=true/1d7feb55-d545-4834-8ad5-d54cafb31dbd.jpeg" /><hr />Cicaloo 1 Frame node supportAs of 26.08, two additional workflows using "Cicaloo 1 Frame node" have been added.Speed up times on my 5090 are about 5 times faster for a 3840x2176 image generation at 55 seconds vs 5 minutes+. My recommendation is to use the "Default workflow" up to 2560x1440 for best quality and the "Cicaloo 1 Frame workflow" for renders up to 3840x2176 higher resolution.Comparison 4K:JXL_IMGEN_DEF_V4b_single_lora workflow<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/e26c98ab-ea62-4852-9ef7-bd450ff9c55c/original=true/e26c98ab-ea62-4852-9ef7-bd450ff9c55c.jpeg" />0.9 lora strength - cfg 1 - 5 minutes 53 secondsJXL_IMGEN_1FRAME_T2I_v4 workflow (CICALOO 1 Frame node - based)<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/5cc178af-9098-4a50-aeb9-72cf6c8c0f52/original=true/5cc178af-9098-4a50-aeb9-72cf6c8c0f52.jpeg" />1.5 lora str / cfg 1.5 - 55 secondsNote: Adjusting lora strength and CFG can have an effect on sharpness versus stability.Prompt: ph0t0r34l, A twenty four year old blonde female looking out a rain-streaked train window, blurred landscape passing by, soft interior lighting, introspective and melancholic , depth of fieldPro tip: The 1 frame gen workflow works quite well at 3840x2176. If you can afford it, try running it at 40 steps for best results.<hr />Authors commentFeel free to post your safe for work results on this LoRa's page too!Please enjoy making content and experimenting with this photorealistic lora and workflow!<hr />Usage:<ul><li>Preferably, use the following trigger word to activate the image generation<ul><li>ph0t0r34l</li></ul></li><li>Seems to work very well on resolutions up to 2560x1440 resolution (Default workflow) and 4K when using the Cicaloo workflow.</li><li>The lora when run at lower lora strength can also help bring out any of your character loras n a more static position.</li><li>As an experiment, you could try increasing/decreasing lora strength and maybe combining both loras with different strengths. 1.0 lora strength is standard for "default" ans 1.5 for the Cicaloo workflow.</li><li>Let me know your feedback, as I believe this is the only photorealistic image generator that exists for H3 so far :)</li></ul>Known issues:<ul><li>The workflow might complain about an image needed to be loaded, just feed it an image (ignore that it will have any effect though)</li><li>You may need to install some frequently used nodes, mostly Kijai's nodes and VideoHelperSuite<ul><li><a target="_blank" rel="ugc" href="https://github.com/kijai/ComfyUI-KJNodes">https://github.com/kijai/ComfyUI-KJNode</a></li><li>When using the Cicaloo workflows, make sure to clome the custom node and download the special VAE, no turbo lora required.</li><li>https://github.com/cicalooo/ComfyUI-MM-1Frame</li></ul></li></ul><hr />Comparisons:Left is with lora, right is without lora.Right click, open image in new tab, notice the difference in sharpness and contrast.<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/4e40ae78-3c5a-48ca-88b4-41b75720b359/original=true/4e40ae78-3c5a-48ca-88b4-41b75720b359.jpeg" /><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/dfde2241-5aba-46b8-9bc3-47f4f0391fd9/original=true/dfde2241-5aba-46b8-9bc3-47f4f0391fd9.jpeg" /><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/0bebedd4-704a-4691-9e7d-77ccf06e4482/original=true/0bebedd4-704a-4691-9e7d-77ccf06e4482.jpeg" /><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/7c97ca95-1578-415b-81bf-cf33a2b25de7/original=true/7c97ca95-1578-415b-81bf-cf33a2b25de7.jpeg" /><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/8ecd6d15-0b57-4dff-a54d-eb6c94acd579/original=true/8ecd6d15-0b57-4dff-a54d-eb6c94acd579.jpeg" /><hr />

Minimax H3 Photorealistic Image Generator (LORA + Workflow)

JXL_TI2V_ImageGenerator_v2.json

JXL_TI2V_ImageGenerator_v1b.json

h3_photoreal_ph0t0r34l_1024-step00003780.safetensors

JXL_H3_TIRV2_V7d_Workflow.zip

JXL_H3_TIR2V_V7G.zip

JXL_H3_TIR2V_V7e.zip

JXL_IMGEN_WFS_4b_update.zip

h3_photoreal_ph0t0r34l_1024-step00003990.safetensors

v1.0

Files with SHA256: f5d9ac2e4219685df33d88ae19359638b32ffdbf682a5af08fd3ed92ac26c5e4