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
19 shown of 41,720
SD3.5 Large Canny-guided images
Generate new images from a prompt while following edges extracted from a reference image. Outputs include the generated image batch and one reference edge-map preview. For a first run, upload a reference image and replace the example prompt with your desired subject, colors, setting, and style. Keep output and reference dimensions at 1024 × 1024, Canny thresholds at 0.3 and 0.6, control strength at 0.66, and denoise at 1. The fixed SD3.5 Large checkpoint and matching Canny ControlNet are used together. Control strength sets how strongly reference edges influence generation. Set it to 0 for generation without edge influence; the reference upload is still required and its edge map is still returned. Guidance start and end are fractions of the denoising schedule from 0 to 1. Keep start at or below end. Reference resizing and generated output dimensions are independent. Match their aspect ratios for consistent framing. Center crop can remove content near the reference borders; disabled cropping resizes to the requested shape and can stretch the image. Reference interpolation changes resizing before edge detection, not generated output dimensions. Inspect the edge preview when changing crop, interpolation, or thresholds. Keep the low threshold below the high threshold; lower thresholds retain more faint detail, while higher thresholds select stronger contours. Variations sets the number of generated images using the same prompt and edge map. Sampling offers seed, steps, prompt guidance scale (CFG), sampler, scheduler, and denoise. Seed -1 selects a fresh random seed each run; use a fixed seed to compare settings with the same initial noise. Begin with the supplied sampling defaults. Denoise 1 uses the full generation schedule; lower values shorten denoising from the empty latent. The reference supplies edges rather than an initial image latent, so this tool does not preserve uploaded pixels. Uploaded transparency is not carried through the connected image path. Negative conditioning is zeroed internally; there is no editable negative prompt.
Anima image generation
Generate square images from text with Anima, using standard sampling or Turbo mode. No source image is required; the result is an image batch returned through Sweet Tea's preview transport, without permanent ComfyUI image saving. For a first run, replace the example positive prompt with your subject, composition, lighting and style. Review the negative prompt for attributes you want to exclude. Leave image count at 1 and Turbo mode disabled, then run with the supplied sampling defaults. Seed -1 produces fresh randomization on each run; use a fixed seed to compare changes with the same initial noise. Output size is fixed to the source's square, approximately 1-megapixel setting, about 1024 × 1024 pixels with dimensions rounded to multiples of 16. This interface has no resolution control. Image count determines the number of returned images; larger batches require more memory. Turbo mode selects the supplied Anima Turbo LoRA together with separate Turbo steps and guidance. Standard defaults are 30 steps and CFG 4; Turbo defaults are 8 steps and CFG 1. Only the active mode's steps and guidance affect sampling. Turbo LoRA strength affects the selected model only in Turbo mode. At CFG 1, negative conditioning does not influence ordinary classifier-free guidance. Advanced controls include both modes' steps and guidance, Turbo LoRA strength, seed, sampler, scheduler and denoise strength. Keep Euler, the simple scheduler and denoise strength 1 for the first run. Steps must be within 1–10000 and CFG within 0–100, even though their primitive input fields accept broader values. Lower denoise strength shortens the noise schedule; this text-to-image workflow has no source image to preserve. The runtime needs these source-specified model files in the corresponding ComfyUI model directories: [anima-base-v1.0.safetensors](https://huggingface.co/circlestone-labs/Anima/resolve/main/split_files/diffusion_models/anima-base-v1.0.safetensors) in models/diffusion_models; [anima-turbo-lora-v0.2.safetensors](https://huggingface.co/circlestone-labs/Anima-Official-LoRAs/resolve/main/anima-turbo-lora-v0.2.safetensors) in models/loras; [qwen_3_06b_base.safetensors](https://huggingface.co/circlestone-labs/Anima/resolve/main/split_files/text_encoders/qwen_3_06b_base.safetensors) in models/text_encoders; and [qwen_image_vae.safetensors](https://huggingface.co/circlestone-labs/Anima/resolve/main/split_files/vae/qwen_image_vae.safetensors) in models/vae. The selected output also requires [Sweet Tea nodes](https://github.com/tea-time-labs/sweet-tea-nodes). Standard and Turbo share the same image output.
Boogu Image Turbo
Generate an image or batch of images from a text prompt with the fixed Boogu Image Turbo model. No source image is required. For a first run, replace the example prompt with your subject, composition, lighting and style. Keep the square aspect ratio, 1-megapixel resolution and one image per run, then generate. The source sampling preset is 4 steps, CFG 1, lcm sampler, sgm_uniform scheduler and denoise 1. Aspect ratio and megapixels jointly determine the output dimensions. Megapixels measures total image resolution; the supplied size guidance describes a square 1 MP image as approximately 1024 × 1024 pixels. Images per run controls batch size. Larger resolutions and batches require more memory. Advanced Sampling exposes the seed, steps, prompt guidance, sampler, scheduler and denoise amount. Seed -1 selects a new random seed each run; use a fixed seed when comparing settings. Lower denoise values shorten the denoising schedule, but there is no source image whose structure can be preserved. Negative conditioning is zeroed internally, so there is no separate negative prompt. This tool does not provide image editing or model switching. Runtime requires boogu_image_turbo_fp8_scaled.safetensors in diffusion_models, qwen3vl_8b_fp8_scaled.safetensors in text_encoders, and ae.safetensors in vae, plus the Sweet Tea output node pack. The supplied dependency inventory marks these assets as missing; they must be available in the execution environment.
Chroma Radiance text to image
Create an image or batch of images from text using the fixed Chroma Radiance x0 model and T5 text encoder. No uploaded image is required; generated images are returned as the Pipe result. For a first run, replace the positive prompt with your subject, composition, lighting and desired style. Keep or edit the negative prompt to describe unwanted qualities, then run at the default 1024 × 1024 size, one image, 30 steps and CFG 3.5. The supplied miniature-bakers prompt is an example; its mention of 8K does not change output dimensions. Width and height set output dimensions in pixels and must be multiples of 16. Images per run sets the batch count. Larger dimensions and batches require more memory. Seed -1 randomizes each run; use a fixed nonnegative seed to reuse starting noise when comparing settings. Advanced controls include prompt guidance (CFG), steps, sampler, flow shift and two independent beta-schedule shape parameters. CFG controls the positive-versus-negative guidance strength. The scheduler is fixed to a beta schedule; Schedule alpha and Schedule beta jointly reshape its sampling-time distribution. The supplied sampling defaults are CFG 3.5, 30 steps, Euler, flow shift 1 and alpha and beta 0.4. Radiance rendering options control internal NeRF tiling and their active sigma range. NeRF tile size -1 uses the model default of 32; 0 disables tiling and may require substantially more memory. Radiance start sigma and Radiance end sigma specify the first and last sigma at which these options apply. Their defaults of 1 and 0 cover the full sampling range. Tile size does not determine output dimensions. The model and text encoder are fixed to chroma-radiance-x0.safetensors and t5xxl_fp8_e4m3fn_scaled.safetensors. This workflow generates images from scratch and provides no image editing, reference-image conditioning or upscale stage.
Photo to whole-body pose map
Upload a photo to produce a standalone pose map and an inspection preview blending that map with the resized photo and detected person boxes. Both outputs belong to the same pose extraction run; the preview is not a separate detection tool. For a first run, keep the 1024-pixel longer edge and the default drawing settings. The source default processes at most one person, choosing detections by descending confidence. Increase Maximum people for a group photo. Detection threshold controls which person boxes reach pose extraction; Keypoint threshold separately controls which estimated points are drawn. Lower thresholds can reveal uncertain detections or points, while higher thresholds can omit people or parts of a pose. The longer-edge setting drives both image resize branches and preserves aspect ratio. Choose which body regions to draw, then tune line width and face point size if needed. Preview blend and blend mode affect only the inspection preview, not the standalone pose map. This workflow produces rendered images, not a saved keypoint data file, and does not preserve uploaded transparency through its RGB image path. Required models are sdpose_wholebody_fp16.safetensors in checkpoints and rt_detr_v4-x-hgnet_fp16.safetensors in diffusion_models. Their supplied installation sources are https://huggingface.co/Comfy-Org/SDPose/resolve/main/checkpoints/sdpose_wholebody_fp16.safetensors and https://huggingface.co/Comfy-Org/SDPose/resolve/main/diffusion_models/rt_detr_v4-x-hgnet_fp16.safetensors.
Realisian Negative Embedding
This Negative Embedding is meant to be used with Realisian models but it can work on other realistic models (it has NOT been tested on anime models).
Corneo's Makima (Chainsaw Man) TI Embedding
This is a textual inversion embedding for Makima from Chainsaw Man. Chainsaw Man is an incredible series that you should try out if you haven't already.
SeedVR2 image upscale
Restore and resize an uploaded image with SeedVR2, producing one output image. For a first run, upload your source image and keep the supplied scale factor of 4, Lanczos resampling, denoise of 1 and color matching set to none. Scale factor multiplies both width and height while preserving aspect ratio. A value of 4 produces roughly sixteen times as many pixels and increases processing demand. Values above 1 enlarge, 1 retains the source dimensions, and values below 1 reduce. Resampling controls interpolation before restoration: Lanczos is the supplied enlargement default, area supports reduction, and nearest-exact retains hard pixel boundaries during resizing. Color matching compares the restored result with the resized source. lab transfers color in CIELAB space; wavelet transfers low-frequency color while retaining generated fine detail; adain matches global per-channel mean and standard deviation. none skips color transfer while retaining geometry alignment. Advanced restoration settings include seed, denoise, steps, guidance scale, sampler and scheduler. Use seed -1 for a fresh random seed each run, or a fixed value to reuse the noise seed. Lower denoise values change the restoration schedule and may retain more source structure. Start with the source baseline of denoise 1, one step, guidance scale 1, Euler sampling and simple scheduling; other sampling configurations are experimental for this tool. Advanced input and output tiling controls separately set spatial encoding and decoding tile sizes and overlaps in pixels. Keep each overlap below its corresponding tile size. The supplied settings are 512-pixel tiles with 128-pixel overlap for both stages. The workflow uses the fixed SeedVR2 7B INT8 model and matching VAE, with conditioning derived from the image rather than an editable text prompt. Generated details may differ from the original; inspect the result when exact fidelity matters.
SDXL two-reference image generation
Create new images with SDXL Base using a positive prompt and visual conditioning from two uploaded reference images. Both references guide the same generated result. For a first run, upload both references, describe the desired subject and style, and keep the 1024 × 1024 output size, reference strengths of 0.75, and denoise amount of 1. Use the negative prompt to describe unwanted features. Image count sets the number of images generated per run with the same prompts and references. Adjust each reference strength independently to change its contribution. Zero removes that reference's conditioning influence; setting both strengths to zero leaves text conditioning. Both uploads remain required in every case. Reference noise augmentation perturbs the corresponding image embedding rather than adding visible grain to the result. The references guide generation through image embeddings. Sampling starts from an empty latent; this is not pixel editing and does not preserve exact reference layouts or uploaded transparency. Center cropping can omit content near reference-image edges. Under advanced reference framing, choose none to avoid center cropping. Under advanced Sampling, use seed -1 for fresh randomization or a fixed seed to compare settings with the same initial sampling noise. Steps controls the number of denoising steps, CFG controls conditioning guidance, and sampler and scheduler select the sampling algorithm and noise schedule. Denoise controls the sampling denoise amount; lowering it does not preserve reference pixels because neither reference supplies the starting latent. Larger dimensions and image counts require more memory. The workflow uses the fixed sd_xl_base_1.0.safetensors checkpoint and clip_vision_g.safetensors vision encoder.
FLUX.2 Klein two-image editing
Generate edited images from a source image, a second visual reference and written instructions using FLUX.2 Klein 9B. For a first run, upload both required images: the main image as Figure 1 and the visual reference as Figure 2. Replace the example instructions with what to change, what to take from Figure 2 and what to retain from Figure 1. For example, request clothing from Figure 2 on the subject in Figure 1 and describe the desired setting. Keep the supplied defaults of 1 output megapixel, 1 reference megapixel, 1 variant, 4 steps, CFG 1 and Euler sampling, then run to receive an edited image. Output megapixels sets Figure 1's resized area in millions of pixels. Those dimensions determine generation size and scheduling, so changing this control also changes Figure 1's reference resolution. Output proportions follow Figure 1, subject to dimension rounding. Reference megapixels independently sets Figure 2's working resolution without setting output size. Each image has its own resize interpolation control. Advanced controls let you request a batch with Variants, choose sampling steps and algorithm, and adjust CFG between the instructed and zero-text conditioning branches. Both branches carry both image references. Seed -1 selects fresh sampling noise each run; use a fixed seed to compare changes with the same initial noise. Both images condition a newly generated result. Exact preservation of pose, identity or unchanged pixels is not guaranteed. This tool has no mask, denoise control or editable negative prompt, and uploaded transparency is not retained through the RGB generation path. Results are returned to Sweet Tea for preview rather than saved by the original ComfyUI SaveImage node.
Seedance image to video
Animate a starting image into one video with prompted action and camera movement, with optional generated audio on Seedance 1.5 Pro. Upload the required image to use as the first frame, then replace the example prompt with action, camera movement and atmosphere suited to your image. For a first run, keep Seedance 1.5 Pro, 480p resolution, adaptive aspect ratio and the five-second duration. Run the Pipe to receive the generated video in Sweet Tea. Duration is measured in seconds and supports 3–12 seconds. The example requests four shots, dialogue and ambience; simplify it or increase the duration to allow more room for the action. Prompted cuts, dialogue and camera effects are creative directions rather than guaranteed timing. Choose an explicit aspect ratio when you need a particular frame shape. Resolution offers 480p, 720p and 1080p, subject to the selected model's support. Model selects among the supplied Seedance 1.5 Pro, 1.0 Pro, 1.0 Lite and 1.0 Pro Fast choices. Generate audio applies only to Seedance 1.5 Pro; the other choices ignore it. With audio enabled, describe desired dialogue and background sounds in the prompt. Fixed camera adds a stationary-camera instruction but does not guarantee a motionless viewpoint; remove conflicting pans, dollies and other camera movement from your prompt when using it. Seed -1 selects a new random value each run; a fixed seed controls generation randomization. AI watermark requests an AI-generated watermark on the video. The uploaded image's alpha mask is unused, so this Pipe does not preserve uploaded transparency. Generation uses the ByteDance provider and involves transferring the input image and prompt to that provider and a provider-priced generation call.
Bad_NegAnatomy Textual Inversion
his is a Textual Inversion I have created, this is mainly useful to improve your overall aesthetics and character anatomy.
Cherry Nobody
Follow me to make sure you see new poses and Nobodys when I post them. Things move fast on this site, it's easy to miss.
Recraft image generation from style references
Generate new images from a text prompt using a custom Recraft style created from one to five reference images. The references guide the visual style; the positive prompt describes the subject, scene, and composition. For a first run, upload the primary style reference, replace the example positive prompt with your own description, choose realistic_image for photographic imagery or digital_illustration for illustrated imagery, and keep the default 1024x1024 output size and image count of one. Run the tool to receive the generated image. Additional style references 2–5 are available in the advanced sections and are bypassed by default. Enable each stage you want to use and upload its reference image. Every enabled reference stage needs an image; leave unused stages bypassed. Use references with a consistent visual style. The negative prompt optionally describes unwanted elements. Output size specifies dimensions in pixels; use a size supported by Recraft. Image count requests one to six images using the same prompt and custom style. This workflow uses the Recraft service and creates a custom style before generating images. It does not expose an existing-style-ID mode or a standalone style-ID output. The references guide style rather than serving as images to edit. Seed -1 randomizes the rerun seed; a fixed seed controls rerun behavior but does not make Recraft results reproducible.
Photography Tools By Stable Yogi (Sd1.5)
Loving this model? Get the Pro version of my models here for exclusive perks and early access to unique resources.
Qwen 2511 image edit
Edit a source image with a written instruction using Qwen Image Edit 2511. For a first run, upload an image, describe the changes you want and the features to retain, and keep the supplied defaults: denoise strength 1, 40 sampling steps, CFG guidance 4, Euler sampler and Simple scheduler. The example instruction requests anime styling while retaining identity. Leave the negative prompt empty unless you want to describe attributes to discourage. Seed -1 randomizes the sampling noise. The image is automatically resized before editing. The resized RGB image supplies both positive and negative reference conditioning and is encoded as the initial sampling latent. Lower denoise values retain more of that initial image's structure. CFG guidance adjusts prompt conditioning separately from denoise strength. A fixed nonnegative seed reuses the noise setting; changing the image, instruction or sampling controls can still change the result. Output dimensions come from the automatically scaled source image. This tool has no manual output-size control, text-only mode, mask-based editing input or separate output-upscaling stage. Uploaded alpha transparency is not preserved because the loader's alpha-mask output is not connected. The selected output returns the decoded edited image. Advanced settings include negative prompting, guidance normalization strength, normalization before CFG and sampling shift. Normalization defaults to strength 1 with pre-CFG normalization off, using post-CFG attenuation. Enabling pre-CFG normalization rescales noise before CFG combination without clamping and can amplify it. Sampling shift defaults to 3.1 and changes the model's sampling configuration independently of denoise and CFG. The workflow uses qwen_image_edit_2511_bf16.safetensors, qwen_2.5_vl_7b_fp8_scaled.safetensors as its Qwen image text encoder, and qwen_image_vae.safetensors.
WaveSpeed image upscale
Upload a source image, choose an upscale model and target resolution, and run to produce the upscaled image through WaveSpeed. For a first run, keep SeedVR2 and the 2K preset; try Ultimate to compare the alternative model, or select 4K or 8K for a larger target. Resolution choices are provider presets, not independently specified width and height. This workflow provides no prompt or detail-strength controls. Uploaded transparency is not preserved through the connected RGB image path.
nfixer for Illuminati Diffusion v1.1
This negative embedding is intended for use with Illuminati Diffusion v1.1, feel free to use it elsewhere but I can't guarantee your success.
Wan 2.7 Reference to Video
Generate a short video featuring subjects from two uploaded reference images using Wan 2.7 Reference to Video. For a first run, upload both required images, then rewrite the example prompt to describe your subjects, their actions, the scene, and camera movement. The source recommends identifiers such as character1 and character2 for reference characters. Keep the default 5-second duration, 720P resolution, and 16:9 aspect ratio initially. The images supply subject references; they are not designated start and end frames. Choose a duration from 2 to 10 whole seconds, 720P or 1080P resolution, and one of the supported landscape, portrait, or square aspect ratios. Use the negative prompt to describe unwanted visual elements. Its supplied default is in Chinese and discourages blur, static scenes, anatomical defects, and text; you can edit it to match your needs. Seed -1 requests a fresh random seed for each run; a fixed seed requests a particular generation variation. Enable Watermark to request the provider's AI-generated watermark. Advanced video output settings offer automatic or explicit container and codec selection; leave both on auto for your first run. This Pipe uses a hosted API node and requires configured provider access. It returns one generated video. The selected graph has two reference-image inputs, no reference-video input, and no separate motion-strength control; describe desired movement in the prompt.
