This is a single file download alternative to the exisiting upload here on civitai. I'll skip the technical mumbo-jumbo (you can read about that <a target="_blank" rel="ugc" href="https://huggingface.co/stabilityai/stable-cascade">here</a> and <a target="_blank" rel="ugc" href="https://huggingface.co/stabilityai/stable-cascade">here</a>) to get straight to the important things like:Which files do i need? As explained more in detail <a target="_blank" rel="ugc" href="https://github.com/Stability-AI/StableCascade/tree/master/models">here</a> you basically need the Stage A model, the EfficientNet encoder model, the Previewer and any combination of the available Stage B and C models. Both are available in large and lite variants (due to site restrictions labeled full and pruned in this case, the file names are the original ones, though), so test and choose what works best on your hardware. The FP16 fixed models were created by KBlueLeaf on <a target="_blank" rel="ugc" href="https://huggingface.co/KBlueLeaf/Stable-Cascade-FP16-fixed">HF</a>.How do i use these with ComfyUI? You can already find a few workflows on civitai (-&gt;Suggested Resources) and the main developer of ComfyUI has provided a starter workflow <a target="_blank" rel="ugc" href="https://gist.github.com/comfyanonymous/0f09119a342d0dd825bb2d99d19b781c">here</a>. You can also find the correct locations for all the different models involved there: Stage A goes into /models/vae/, Stage B and C into /models/unet/, the text encoder into /models/clip/. StabilityAI provides a Stage B and C model specifically for ComfyUI users, which i also reuploaded here, no idea, though, in what way they exactly differ.How do i use these with Automatic1111/Forge? At the moment i only know of this <a target="_blank" rel="ugc" href="https://github.com/blue-pen5805/sdweb-easy-stablecascade-diffusers">extension</a> implemention of Stable Cascade for the Automatic/Forge platform. Let's hope native support is on its way.DISCLAIMER: As always, if you are the owner of the model and don't want to see it uploaded here, then please do reach out to us and we will find a solution.

Stable Cascade by Stability.AI (single files)

Original Repository on huggingface: <a target="_blank" rel="ugc" href="https://huggingface.co/stabilityai/stable-cascade/tree/main">https://huggingface.co/stabilityai/stable-cascade/tree/main</a>

stage_a.safetensors

Stage A

TextEnc

EffNet

Previewer

Stage C (for ComfyUI)

Stage B (for ComfyUI)

Stage C

Stage B

<h3 id="-b44gcqft9"></h3>===========================================================================RES-M - Think of this as a flip-side of A. Clearer and merged from more capable counterparts of models to come, including clearer anatomy and concepts. Thanks for all the support, more to come. Please share your generations!RES-X - A few editions coming soon now that FLUX is a primary focus. Plan to release the "final contenders" in the coming week or so (9/14)RES-A - The 'A' is for analog. It's a richer, softer, more photo-realistic take on Stable CascadeR35 models are created by DR34MSC4PE with all the trappings you've come to expect as well as some cool bonuses:<ul><li>Generate stunning, enormous images quickly or "Analog" poloroids in a snap. The model's compression dial allows for whichever poison you prefer!</li><li>NSFW tuned and capable of realism and artistic/anime</li><li>Full BF16 C/B stages with functional workflows allow for quick, easy setup in Comfy UI. No managing unets/CLIP!<ul><li>Users who prefer Diffusers version of the models can claim them in the coming weeks!</li></ul></li><li>Comfy nodes (like those found in some of the example images) that affect CLIP for SDXL work wonderously on Cascade as well. LORA support demonstrated and disabled therein.</li><li>Support for proper use in Automatic/SDNEXT - Stay Tuned!</li></ul>Limitations:<ul><li>Mid range detail, eyes can be inconsistent/different-colors without adjusting resolution/compression/steps accordingly</li><li>Can gravitate toward animated output at low step/compression</li><li>Nude male anatomy (WIP)</li><li>Blurryness in some NSFW concepts at low step counts (WIP)</li></ul>===========================================================================<h3 id="dr34msc4pe-is-q7m80xs6n">DR34MSC4PE is</h3>c0ur4ge (trainer/qa/inference code) /eraser851 (captions/tooling code/data/qa) /===========================================================================<h2 id="usage-guide-for-all-r35on4nce-variants:-tsjo8qol7">Usage Guide - For all R35ON4NCE Variants:</h2><h3 id="use-with-the-default-cascade-workflow-or-the-advanced-found-in-the-demo-images.-oc6xw7ijb">Use with the default Cascade Workflow or the advanced found in the demo images.</h3>Examples (pastebin):<ul><li><a target="_blank" rel="ugc" href="https://pastebin.com/rMJw716p">Basic</a></li><li><a target="_blank" rel="ugc" href="https://pastebin.com/LrHWvbf9">R35 + SDXL Prompter + 2x Basic Upscale</a></li><li><a target="_blank" rel="ugc" href="https://pastebin.com/uf9YaL1L">R35 to SDXL Pipeline (B-Stage Bypass)</a></li></ul>===========================================================================<h2 id="c-stage-g97warpmd">C-Stage</h2>===========================================================================CFG: 3-8Latent Compression: 22-42Steps: &gt;20-80Note: higher steps can reduce blur in mid-range details in conjunction with compression settings. The interplay between these settings is the difference between mush and miracles.CFG works differently in Cascade. While it can improve prompt coherence it can impact (or improve) image quality in unexpected ways. Lower CFG can often yield more detailed images, for example.===========================================================================

R35ON4NCE

This is the A-stage (VQGAN VAE) component of Stable Cascade. It is packaged in with the Comfy C-stage checkpoints automatically and does not need to be download except for UNET style inference or within the advanced workflows. It produces crisper results then the A-stage checkpoint released with Cascade after fine tuning on HQ images by <a rel="ugc" href="https://huggingface.co/madebyollin/stage-a-ft-hq">madebyollin</a>.

R35-Stage_A_v1.0.safetensors

A-Stage_Universal

R35-M_Cstage_bf16

B-Stage_Universal

R35-Analog_Cstage_bf16

B-Stage_v1.0

UltraPixel

This is the implementation for UltraPixel: Advancing Ultra-High-Resolution Image Synthesis to New Peaks

stage_a

stage_b_lite_bf

stage_c_bf16

Files with SHA256: 0a3f5a0b618494f81964dc719dbaaa9bd52f207d464d5fbf45442d363e1c0b5a