Resource directory
Find what you need for your workflow.
Guides: techniques, setup, and troubleshootingSearch Sweet Tea Pipes, external workflows, models, node packs, and guides in one evidence-linked catalog.
Start with a workflow instead ↗Search results
All resources
17 shown of 35,030
Qwen-Image 2512 edge-guided generation
Generate new images from a prompt while following an uploaded reference image's Canny edges. The Pipe returns generated images and an edge-map preview. For a first run, upload a reference and replace the example positive prompt with your desired subject, appearance, lighting and style. Review the negative prompt for unwanted attributes. Keep output size at 1.6 megapixels, control strength at 1 and Lightning mode off, then run. Output dimensions follow the resized reference's aspect ratio. The reference supplies edges and dimensions; generation starts from an empty latent and does not directly preserve source colors or texture. Uploaded transparency is not retained. Lower control strength for more freedom, or set it to zero to remove edge conditioning while retaining the reference-derived dimensions and preview. The Canny thresholds are normalized values: keep the low threshold at or below the high threshold. Lower thresholds admit more edges; higher thresholds focus on stronger boundaries. Inspect the returned edge map when changing them. Control start and end are fractions of denoising progress from 0 to 1; keep start below end. Lightning mode selects the fixed four-step LoRA and separate sampling controls. Its supplied defaults are 4 steps and CFG 1; standard mode uses 50 steps and CFG 4. Only the selected mode's steps and CFG apply. Use positive step counts; the downstream sampler supports 1–10000 steps and CFG 0–100. Euler and the simple scheduler are the supplied choices for both modes. Increase Image count for a batch. Seed -1 randomizes each run; use a fixed seed to compare changes with otherwise matching settings. Advanced controls also expose reference interpolation, denoising amount, sampling shift and Lightning LoRA strength. Denoise defaults to 1; lowering it does not turn this into pixel-preserving image editing. Sampling shift affects both modes. LoRA strength affects Lightning mode only and does not automatically change its steps or CFG. Larger output sizes and batches increase memory use. This interface supports Canny guidance only. Pose or depth guidance would require graph changes.
Flux.2 Klein base 4B image edit
Generate edited images from a reference image and a written instruction using Flux.2 Klein base 4B. For a first run, upload a source image, describe your desired changes and any features to retain, and keep the supplied defaults: 1 megapixel, one image, CFG guidance 5, 20 sampling steps and Euler. The example instruction requests anime styling while retaining identity. Leave the negative prompt empty unless you want to describe attributes to discourage. Seed -1 selects random sampling noise. Output megapixels resizes the reference while following its aspect ratio. The resized image's dimensions also set the generation size. Resize method controls interpolation of that reference before encoding; it does not add a separate upscaling stage. Advanced controls provide image count, negative prompting, CFG guidance, seed, sampling steps and sampler choice. Use a fixed nonnegative seed to reuse the noise setting; changing the reference, instruction or sampling settings can still change the result. CFG combines positive and negative conditioning, with the image reference attached to both branches. The source image supplies reference conditioning, while generation starts from an empty latent. There is no image-to-image denoise-strength control, text-only mode or mask-based editing mode in this tool. Uploaded alpha transparency is not preserved by the connected RGB image path. The output contains generated images. The workflow uses flux-2-klein-base-4b-fp8.safetensors, qwen_3_4b.safetensors as its Flux.2 text encoder, and flux2-vae.safetensors. It uses the base 4B model despite the original subgraph's historical distilled title.
ComfyUI-DepthCrafter-Nodes
A port of tencent/DepthCrafter into ComfyUI
GPT Image generation and editing
Generate one to eight images from a text prompt, or edit an uploaded reference with GPT Image. Returned images appear together in the Pipe result. For a first run, leave Source image disabled, replace the example prompt with a description of your desired image, and keep the default square size and one image. For editing, enable Source image, upload a reference, and describe the desired changes. The upload supplies two inputs: an RGB reference image and a separate mask derived from its alpha channel. Transparent areas become white mask areas designated for replacement. Uploaded transparency is not retained in the RGB reference. There is no separate mask upload or independent mask control; the same upload supplies both inputs. Choose a model and an Output size preset or automatic sizing. Custom width and height apply only when Output size is Custom and Model is GPT Image 2. Each dimension must be a multiple of 16 from 1024 to 3840 pixels. Image count requests one to eight results. Quality affects provider cost and generation time. Background requests automatic, opaque, or transparent provider output; preservation of generated alpha in the returned Pipe images is not established. The provider backend does not implement seed support, so seeds cannot reproduce earlier results. Runs use the partner API and require its access and credits.
ComfyUI-Stream-Pack
A suite of custom ComfyUI nodes for building real-time video and audio workflows using ComfyStream.
DEN_MelkorMancin_Artsyle
I don't think I do a good job with styles. I do something wrong in the text or data set choices, or maybe I just don't use enough consistent images. This was done with circa 100 consistent styled images.
Architectural Magazine Photo Style: Better Interior embedding helper (archmagazine)
Inspired by a medium type tag "Architectural Diggest Photo" which is recognized by Stable Diffusion, I researched architectural magazines further and trained this textual inversion to be used as a better version of...
Realism Render
(embedding:novanrealism3) = render (embedding:novanrealism2) = lighting (embedding:novanrealism1) = QCC
comfyui-gimp-nodes
Redesign of to enable GIMP compatibility
Pastel Style TI
Pastel style was one of my first ever commissioned embeddings and still is useable on newer models. It adds a pastel typical style to the general image. For best use, I would use it with any kind of portrait type image.
sakimi-style
Originally posted to HuggingFace by sd-concepts-library Sakimi Style on Stable Diffusion This is the concept taught to Stable Diffusion via Textual Inversion. You can load this concept into the Stable Conceptualizer...
Style Swamp Magic
This is my sixth publicly released Textual Inversion, called Style-Swampmagic. This model was trained on a single image using DreamArtist. This TI gives things as the name implies, a swampy/earthy feel. Things come...
LTX 2.3 basic GGUF 720p workflow — v1.0
This is same as default WF in ComfyUI, but it uses GGUF custom node. Basically, you can insert images, audio, and video into any frame, so anything is possible.
ComfyUI-Seed-API
ComfyUI nodes for BytePlus foundation models via API
Erin Nobody
Meet Erin Nobody, this person doesn't exist. You may think she looks like someone you know, but she isn't. Just the imprecise nature of diffusion. Most images were generated using the legendary A-Zovya Photoreal and...
Axsens inspired Hypernetwork
This Hypernetwork is only to be used for personal curiosity and to obtain beautiful anime character designs. Please do not sell images made with this.
Leah Vicious (V2)
LEAH VICIOUS (V2) Leah is from Germany and of course part of the gigantic Vicious family. They are nice people and gave birth to a lot of pretty daughters, the best about them is... they are entirely fictional.
