Directorio de recursos
Encuentra lo que necesita tu flujo.
Guides: techniques, setup, and troubleshootingBusca Pipes de Sweet Tea, flujos externos, modelos, paquetes de nodos y guías en un catálogo con evidencias.
Empieza con un flujo ↗Resultados de búsqueda
Todos los recursos
20 mostrados of 41.761
Inflate an image subject
Give a subject in an uploaded image an inflated appearance using Qwen Image Edit 2511 and the INFL8 LoRA. This Pipe returns one edited image. For a first run, upload your source image and replace the example Edit instruction with a subject visible in it, such as “inflate the chair.” Keep Inflation effect enabled, LoRA strength 1, denoise 1, 40 steps, CFG 4, Euler sampler and simple scheduler. Leave the negative prompt empty unless you have attributes to discourage. Seed -1 requests a fresh random variation each run; use a fixed seed to compare settings with consistent starting noise. The source image is automatically scaled and supplies both image conditioning and the starting latent. There are no independent output-size controls. Denoise ranges from 0 to 1: lower values retain more starting-image structure but may weaken the edit. LoRA strength changes INFL8's influence rather than specifying an exact physical expansion. Bypass Inflation effect to use the same image-editing path without INFL8. LoRA strength has no effect in this mode, and inflation behavior depends on the base model and instruction. The negative prompt remains available to discourage unwanted attributes. Sampling controls select the step count, prompt guidance, denoising algorithm and noise schedule. Advanced guidance normalization controls adjust its strength and whether normalization occurs before CFG combination; the default false setting uses post-CFG attenuation, while true uses unclamped normalization that can amplify the prediction. Sampling shift adjusts the model's noise schedule and defaults to 3.1. Keep these source defaults for your first run. This is a generative whole-image edit without a mask input, so surrounding details may also change. Uploaded transparency is not preserved. The result is returned through Sweet Tea's image preview output; the workflow does not automatically save it through ComfyUI SaveImage.
Chromatic aberration
Create an image with chromatic-aberration color fringing from an uploaded source image. For a first run, upload your image and run with the saved settings: Amount 31 and Pattern Linear. These settings apply horizontal color separation rather than a neutral preview. Set Amount to 0 to preview the loaded image without the effect, then change the amount or pattern and run again. Amount uses a unitless 0 to 100 scale, not pixels or a percentage. Higher values increase the displacement of red and blue samples in opposite directions, while green stays at its original position. The same amount produces different displacement across patterns. Choose Linear for a uniform horizontal shift. Radial produces separation along directions radiating from the image center, increasing toward the edges. Barrel increases separation with the square of distance from the center; it separates color channels rather than warping the whole image. Swirl shifts perpendicular to the radial direction, producing separation around the center. Diagonal shifts along both image axes in normalized image coordinates, so its apparent angle depends on the image proportions. If Pattern appears as a text field, enter Linear, Radial, Barrel, Swirl or Diagonal exactly. Radial, Barrel and Swirl account for the source aspect ratio and have no displacement at the center. Amount 0 disables every pattern. Strong settings sample farther from each pixel and can produce pronounced edge artifacts. Output dimensions follow the source image. Uploaded transparency is not preserved: the image loader's separate alpha mask is unused, even though the shader copies its input texture's alpha. This tool applies only chromatic aberration; the other image adjustments in this family are separate tools.
Extend video with Vidu
Generate a video continuation from an uploaded clip with Vidu Q2. For a first run, upload the required source video, describe what should happen next, and keep Q2 Pro, a 4-second extension, and 720p resolution. The continuation prompt can include motion or camera direction and accepts up to 2,000 characters. Leave it blank to request a continuation without text guidance. Access to the authenticated Vidu partner API is required. Choose Q2 Pro or Q2 Turbo, an extension duration of 1–7 seconds, and 720p or 1080p output. Duration specifies the generated continuation rather than the uploaded clip's length. Keep seed -1 for a new random seed each run, or enter a fixed seed; identical provider results are not guaranteed. End frame guidance is enabled by default under advanced settings. It derives an ending target from the end of the uploaded video, which can be useful when requesting a return toward that image. Bypass the stage to request extension without an ending image. There is no separate end-frame upload, and a seamless loop is not guaranteed. Advanced video output settings provide automatic or explicit file type and codec selection. Keep both on auto initially. These settings affect output encoding rather than the generated scene. The selected result is Vidu's returned video; this graph has no separate concatenation or source-audio restoration stage, so do not assume that the original clip or its audio is preserved in the result.
Qwen Image Edit 2509
Edit an uploaded image with natural-language instructions and receive an edited RGB image in Sweet Tea. For a first run, upload your source image, replace the example instruction with the change you want, describe details to retain, and leave Lightning mode and the default sampling settings enabled. Lightning mode selects the supplied acceleration LoRA and the Lightning steps and CFG settings, initially 4 and 1. Disable it to use the base model and the separate standard settings, initially 20 steps and CFG 4. Each mode retains its own values; inactive-mode settings take effect when you switch to that mode. Keep Euler and the simple scheduler for the source configuration. The configured Qwen editing model, text encoder and VAE are required, along with the Lightning adapter for Lightning mode. Denoise ranges from 0 to 1: lower values retain more source structure, while 1 permits stronger changes. Seed -1 chooses a new random variation each run; use a fixed seed to compare settings with the same initial noise. Negative prompt describes unwanted content. At the default Lightning CFG of 1, negative conditioning has no ordinary classifier-free guidance effect; use standard mode with CFG above 1 when negative guidance is needed. Both modes expose their own steps and CFG controls. The downstream sampler accepts 1–10000 steps and CFG from 0 to 100. Advanced settings also expose Lightning adapter strength, guidance normalization strength, normalization before CFG, and sampling shift. Adapter strength affects Lightning mode only; the normalization and sampling patches apply in both modes. Keep their source defaults for a first run: adapter strength 1, normalization strength 1, normalization before CFG disabled, and sampling shift 3. Enabling normalization before CFG removes its attenuation-only behavior and can amplify. The source image is automatically scaled before conditioning and latent encoding. There is no manual output-size control or mask input, and the result may have different pixel dimensions from the upload. Uploaded transparency is not preserved. The selected output returns an image preview rather than permanently saving a PNG file.
Vidu Q2 start and end frame video
Generate a short video guided by a start image, an end image, and a description of the motion between them. For a first run, upload both endpoint images and replace the example prompt with the action, camera movement, and scene transition you want. Both images are required; there is no start-frame-only mode. Keep the defaults of Vidu Q2 Pro Fast, 5 seconds, 720p, and small movement initially. Prompts support up to 2000 characters. Duration supports 2–8 seconds in whole-second increments. Resolution offers 720p or 1080p. Movement amplitude sets object movement to small, medium, large, or automatic; describe camera movement in the prompt. Match the requested action to the selected duration. The model selector also offers Vidu Q2 Pro and Turbo. Seed -1 randomizes each run; a fixed seed controls generation randomization. The Pipe returns one generated video. Advanced video output settings offer automatic, MP4, MKV, or WebM file type and automatic, H.264, or AV1 codec selection. Keep both on auto for a first run. Generation sends the endpoint images and prompt to Vidu through a paid provider call. Endpoint images guide a generative transition rather than a deterministic blend.
Character turnaround
Generate one image containing multiple views of a character. For a first run, upload a character reference in Source image, keep the default prompt, 9:16 aspect ratio and 2K resolution, then run. The default requests four close-up face views on a white background. To generate without a reference, bypass Source image and add the character's appearance to the prompt; leaving the upload empty alone does not select text-only mode. For full-body framing, change the prompt to: “A 360 turnaround full body view of the subject, straight in 4 different angles. Front view, half turn, side profile, back view. Full-body shots, all white background.” Set Aspect ratio to 16:9, matching the source full-body branch. Headshot and full-body framing are prompt choices within the same generator; there is no separate preset switch. Aspect ratio controls the complete image, not individual view panels. Auto follows the reference image's ratio when supplied; choose an explicit ratio when generating without a reference if the output shape matters. Resolution offers 1K, 2K and 4K; 2K and 4K use Gemini's native upscaler. Model, Response format and System instructions remain independently editable. System instructions set the provider's foundational behavior. IMAGE+TEXT requests an accompanying provider text response, but this Pipe returns only the image. Seed -1 requests a new random seed each run. Fixed seeds offer best-effort repeatability rather than deterministic results. This tool performs one generation, rather than separately generating and stitching headshot and full-body panels. Requested view counts, pose accuracy and identity consistency are generation instructions rather than guarantees.
ChronoEdit image editing
Create one prompt-guided edited image from an uploaded source using ChronoEdit 14B. The source supplies both visual conditioning and the starting frame. Internally, the workflow generates and decodes a five-frame sequence, then returns only the fifth image; it does not export a video. For a first run, upload the image to edit, replace the example advertising prompt with a description of your desired result, leave the output at 720 × 720 pixels, and keep fast mode off. The negative prompt discourages unwanted attributes; its supplied Chinese text can be edited or replaced with exclusions appropriate to your scene. Standard mode uses the supplied defaults of 20 steps and CFG 4. Fast mode selects the fixed ChronoEdit distillation LoRA and its separate defaults of 4 steps and CFG 1. Standard and fast step counts and guidance affect only their respective modes. LoRA strength affects only fast mode. The mode name describes this distilled configuration; actual runtime depends on the execution environment. Width and height set the output canvas in pixels, in multiples of 16. Changing the aspect ratio can change source framing. Reference crop controls only the visual-conditioning branch: none avoids a center crop, while center uses the middle of the source. It does not crop the separately connected starting-frame branch. In advanced Sampling, use seed -1 for a new random seed each run or a fixed seed to compare changes with the same sampling noise. Sampler, scheduler and denoise affect both modes. Denoise uses a fraction of the noise schedule and is not a direct source-preservation slider in this temporal setup. Keep the supplied sampling values for a first run. Advanced positional encoding and model sampling shift also affect both modes. They adjust model encoding and its noise schedule, rather than directly controlling canvas size, image translation or frame selection. The sequence length and selection of the fifth image remain fixed. The uploaded alpha mask is unused, so source transparency is not preserved. The output is a generated edit, and the source framing or details can change.
Z-Image: Text to image
Generate square images from text using the fixed Z-Image base model and Qwen text encoder. Generated images are returned through Sweet Tea's image preview output. For a first run, replace the example positive prompt with your subject, setting, composition, lighting and visual style. Leave the negative prompt empty unless you want to discourage specific elements. Keep image count at 1, steps at 25, prompt guidance (CFG) at 4, sampler at res_multistep, scheduler at simple and denoising strength at 1, then run the Pipe. No source image is required. The saved resolution setting is square at approximately 1 megapixel, around 1024 × 1024 pixels. Resolution and aspect ratio are fixed in this interface. Image count creates a batch of images; larger batches require more memory. Seed -1 randomizes each run. Use a fixed seed to compare prompt or sampling changes with the same initial noise. Steps control sampling work, and CFG controls prompt guidance; excessive guidance can harm the image. Sampler and scheduler offer alternative sampling behavior. Change one setting at a time when comparing results. Denoising strength ranges from 0 to 1. Keep 1 for the saved empty-latent generation setup; lower values shorten the noise schedule and do not provide image-to-image editing. Advanced noise schedule shift adjusts the model's schedule and starts at 3. The model, text encoder and VAE are fixed to z_image_bf16.safetensors, qwen_3_4b.safetensors and ae.safetensors. This Pipe provides text-to-image generation, with no image upload or selectable model mode. Its preview output does not perform the original workflow's permanent PNG-save operation.
Text-to-Image · Flux.1 Schnell Full: Text to Image
Generate images quickly with Flux.1 Schnell full version. Uses Apache2.0 license, requires only 4 steps to generate images while maintaining good image quality.
Wan 2.2 speech-driven video
Generate a speech-driven video from a reference image, a speech recording and a scene prompt. Separated vocals condition generation; the trimmed original recording supplies the soundtrack. Additional outputs preview the prepared reference, separated vocals and instruments. First run: upload both required files, replace the example positive prompt to match your reference and desired motion, and retain the source defaults: 960 × 528 fill/crop framing, 49 frames per chunk, 5 steps, CFG 1, uni_pc sampling, simple scheduling and 16 FPS. The negative prompt describes unwanted content and also feeds negative-attention guidance. Seeds of -1 randomize each run; use fixed seeds when you want to retain the noise choices. Width, height and framing prepare the reference and determine generation dimensions. Use positive dimensions divisible by 16. Stretch, crop and pad offer different ways to fit the reference; keep-proportion mode can change the resulting dimensions. Frames per chunk follows the 1 + 4n convention, such as 49. It does not choose the number of chunks: the graph generates an initial segment and one continuation, while five further source examples remain bypassed and are not selectable here. Advanced settings retain independent initial and continuation denoising, generation batch size, three stacked LoRA strengths, sampling shift and negative-attention guidance strength, mixing and clipping. Start with their authored defaults. Batch size 1 is the single-video recipe, not a chunk-count setting. The Lightning HIGH and LOW LoRAs are applied in series, followed by Lightx2v; they are not alternative model modes. FPS changes playback timing without generating more frames or retiming audio. The starting frame offset defaults to 2 and changes the final frame selection without trimming the soundtrack. Both controls can affect synchronization. Audio trimming remains fixed at source positions 0–6; the supplied evidence does not establish those positions' units or boundary behavior. Output duration is not automatically matched to the recording. The video output defaults to MP4 with NVENC H.264 and requires a compatible encoding runtime. Model and encoder selections retain the authored recipe. Uploaded transparency is not preserved through the RGB reference-image path.
Flux Klein crop inpaint
Replace a selected region with a Flux 2 Klein generated crop and blend it onto the original RGB image. First run: upload the image, provide a manual mask through the image input's mask workflow, and replace the example Edit instruction with your desired change. Leave Use mask enabled, Automatic mask disabled and Use native crop size enabled. Set Object to select to an object present in your image: SAM2's image output feeds the crop even in manual-mask mode. Run and inspect the edited image, processed mask and input crop preview. The processed mask preview is before cutoff and final feathering, not the final blending mask. For text-detected selection, keep Use mask enabled, enable Automatic mask and name the object to select. Detection threshold, blur, offset and inversion adjust this selection. Disable Use mask for whole-image image-to-image processing. Neither mask switch bypasses SAM2 processing. The crop supplies both the starting latent and a latent reference. Denoise controls change strength throughout that crop; the selected mask controls cropping and final compositing, not a diffusion noise mask. Crop padding adds surrounding context in source-image pixels. Mask coverage adjusts the resized mask, Mask cutoff thresholds it, and Composite edge applies final expansion and feathering. Inverting a processed or composite mask does not change the earlier crop selection. Native crop size uses the crop dimensions as resize targets, with processing dimensions subject to the resize stage's divisibility setting. Disabling it stretches the crop to a square whose width and height equal Sampling resolution in pixels; this is not a maximum-side limit. The graph resizes the generated image back to the crop dimensions but does not explicitly route a correspondingly restored processed mask into compositing. Inspect alignment in this mode. Large native crops can increase memory use. Advanced controls include seed, sampling algorithm, schedule, steps, Flux guidance and resolution-dependent sampling shifts. Seed -1 requests fresh randomization; a fixed seed reuses the noise choice. Desaturation reduces generated-crop saturation before blending, with a selectable grayscale conversion method. The source suggests 0.05–0.20 for saturation mismatch, but this does not guarantee brightness or color matching. Uploaded transparency is not preserved as output alpha.
Hamunaptra Style
This is a Style embedding which turns everything in your image towards an Ancient Egyptian style! Think Pyramids, the river Nile, and Pharaohs! Simply throw Style-Hamunaptra on to the end of your prompt (or the...
FLUX object eraser
Produces an edited image with masked objects removed and the background reconstructed through the BFL FLUX Erase API. No text prompt is required. For a first run, prepare a PNG whose transparent regions cover the unwanted object and any shadow or reflection you want removed. Keep the surrounding areas opaque, and preserve the original RGB colors underneath the transparent regions when saving the PNG. Upload it as Source image, leave Mask expansion at 10 pixels and Seed at -1, then run the tool. The same upload supplies both the RGB image and its alpha-derived removal mask: transparent regions become white mask areas to erase, while opaque regions become black areas to preserve. An entirely opaque image does not specify an erase region. This tool has no separate mask upload or mask-painting control. Mask expansion accepts 0–25 pixels and enlarges the erase region to cover object edges. Increasing it includes more surrounding image; reducing it limits expansion near details you want to keep. Seed -1 randomizes each run; a fixed seed reuses the reconstruction noise setting. The output is a reconstructed RGB image, not a transparency-preserving cutout. Execution sends the image and mask to BFL and requires access to its paid partner API.
FLUX.2 Klein Base 9B text to image
Create an image or a batch of images from text using the fixed FLUX.2 Klein Base 9B model. Describe the subject, composition, lighting and style in the positive prompt; optionally list unwanted elements in the negative prompt. No uploaded image is required. For a first run, keep the 1024 × 1024 size, one image, 20 steps, CFG 5 and Euler sampler. The supplied motorcycle prompt is a ready-to-run example. Width and height jointly set the latent image size and the resolution-aware sampling schedule; use multiples of 16. Larger dimensions and batches increase memory requirements. Advanced controls let you tune guidance, steps, sampler and batch size. Higher CFG increases the influence of prompt conditioning and can change the visual balance; more steps increase sampling work without guaranteeing a better result. Seed -1 randomizes each run; use a fixed seed to reuse the initial noise while comparing other settings. The runtime requires flux-2-klein-base-9b-fp8.safetensors, qwen_3_8b_fp8mixed.safetensors and full_encoder_small_decoder.safetensors. The source graph uses the Base model with its supplied sampling settings; it does not provide an image-editing mode or a selectable distilled-model preset.
Recreate an image with realistic detail
The reference image supplies a generated scene description, rather than direct image conditioning. A style prefix and detail suffix surround that description, and the resulting positive prompt drives both Grok generation and Z-Image Turbo refinement. The default style is black and white photography. Refinement denoise controls how much the generated image changes. Before refinement, the image is resized to the chosen base megapixel count, enlarged fourfold with RealESRGAN, and reduced by half, producing approximately four times the base pixel count. Grok resolution affects the initial generation; base megapixels determines the final refinement size. The negative prompt affects only refinement and becomes useful when refinement CFG is increased above its default of 1.
HiDream O1 image studio
Create an image batch with HiDream O1 Image Dev by editing an uploaded reference or generating from a text description. For a first run, upload a source image, keep Image edit enabled, replace the example Positive prompt with your requested change, leave Refine prompt off, and run with the existing defaults. Results are returned through Sweet Tea's image output; this graph does not save files through ComfyUI SaveImage. An upload is required by this interface in both modes. With Image edit enabled, the original upload supplies reference conditioning. Output width and height follow the upload, rounded down to multiples of 32. Use dimensions of at least 64 pixels and keep the calculated canvas within 4096 pixels per dimension. To calculate a different edit canvas, enable the currently bypassed Resize edit output stage and set Edit megapixels, measured in millions of pixels. Aspect ratio follows the upload. This stage changes canvas sizing only; reference conditioning and prompt refinement still receive the original upload. Resize method affects the intermediate sizing copy, rather than directly filtering the generated result. Disable Image edit for text generation and describe the complete desired image in Positive prompt. Text width and Text height specify pixels, in multiples of 32 from 64 to 4096. Text batch count and Edit batch count independently select the number of results for their respective modes. Text mode does not use the upload as image-generator reference conditioning, but the interface still requires it. Refine prompt optionally uses Gemma 4 and the upload to expand your request into an English visual description, including in text mode. It may add interpretation or inferred details; leave it off when exact wording matters. Refinement tuning controls text sampling, its separate seed and the response token limit. Temperature and token filters apply when Sample refinement text is on. An insufficient token limit or malformed JSON response can prevent prompt extraction. Refinement thinking requests model-supported thinking behavior; false is the source configuration for JSON extraction. The starting image settings are 28 steps, normal scheduler, CFG 1 and Add noise enabled. CFG changes conditioning influence; at 1, conventional classifier-free negative guidance is inactive. Denoise changes the sampling schedule from an empty latent in both modes, rather than directly controlling preservation of uploaded pixels. Advanced Sampling noise tuning exposes the initial and final per-step noise multipliers and clipping threshold. Image Seed and Refinement seed control separate random processes: -1 randomizes each run, while fixed seeds help compare settings. This tool accepts one reference image. Uploaded alpha is not carried through the RGB reference path. Edit canvas dimensions are rounded rather than clamped, so check the resulting size when using a very small image or an extreme aspect ratio.
Grok Imagine image generator
Generate one or more images from a text prompt using Grok Imagine. Describe the subject, composition, lighting, style, and any desired text. No uploaded image is required. For a first run, use the supplied example or replace it with your own prompt. Keep grok-imagine-image-quality, the square 1:1 aspect ratio, 1K resolution, and one image, then run the Pipe. Returned images appear in Sweet Tea results; the output stage does not permanently save files. The model field selects the Grok Imagine model. Aspect ratio sets the width-to-height shape, while resolution selects a size tier supported by that model. Keep the supplied values unless you know another combination is supported; the supplied evidence does not enumerate alternative model, aspect-ratio, or resolution values. Number of images requests 1–10 results from the same prompt. Seed controls whether the provider node re-runs, rather than making images reproducible. Leave it at -1 to randomize on each run. A fixed seed does not guarantee the same image. Generation sends the prompt to the external Grok provider and uses a paid provider call.
Character camera views
Upload a character image and describe the camera view you want. This Pipe uses Qwen Image Edit 2511 with the multiple-angles LoRA to generate one image per run. The uploaded image supplies both reference conditioning and the initial image latent. For a first run, upload your image, keep the close-up instruction and leave Fast mode enabled. Replace the instruction to request another view. The eight source camera instructions are shown below; leading whitespace and trailing line breaks inside the code blocks reproduce the source strings. Close-up: ``` Turn the camera to a close-up. ``` 45 degrees right: ``` Rotate the camera 45 degrees to the right. ``` Aerial: ``` Turn the camera to an aerial view. ``` Wide-angle: ``` Turn the camera to a wide-angle lens. ``` Low-angle: ``` Turn the camera to a low-angle view. ``` 90 degrees right: ``` Rotate the camera 90 degrees to the right. ``` 90 degrees left: ``` Rotate the camera 90 degrees to the left. ``` 45 degrees left: ``` Rotate the camera 45 degrees to the left. ``` Fast mode coordinates the Lightning LoRA, Fast steps and Fast CFG through one source boolean. Its defaults are 4 steps and CFG 1. Disable Fast mode to use the model without Lightning and the independent Standard steps and Standard CFG defaults of 40 and 4. Each mode's tuning fields affect only that mode. Keep Euler and the simple scheduler for the source setup. Negative prompt guidance has little or no effect at CFG 1. Denoise defaults to 1 to allow substantial viewpoint changes. Reducing it retains more of the initial image structure but can prevent the requested camera change. Camera LoRA strength adjusts the multiple-angles adaptation in both modes; Lightning strength applies only in Fast mode. These are generated interpretations of a single image, not measured 3D rotations. Hidden surfaces and character details may be invented or changed, and requested degree values are instructions rather than geometric guarantees. The source automatically scales the input before generation and offers no independent output dimensions. Uploaded transparency is not preserved by this RGB-only path. Run again with another instruction to create additional views.
Color balance
Create a color-graded image by adjusting red, green and blue separately in shadows, midtones and highlights. For a first run, upload a source image and run with the saved settings to preview the source's stylized grade. These defaults include strong shadow and highlight shifts rather than a neutral starting point. To start without a color shift, set all nine color adjustments to 0, then adjust the tonal region you want to change and run again. Each adjustment uses a unitless -100 to 100 scale. Positive red, green or blue values add that channel; negative values shift toward cyan, magenta or yellow respectively. Tonal regions are weighted by the original image's HSL lightness and overlap, so they are not hard selections. The nine adjustments combine in one grade. In advanced controls, Preserve luminosity is enabled by default. It restores original HSL lightness after the shifted RGB values are clamped. Disable it to allow the color shifts to alter lightness too. This preserves HSL lightness, not every measure of perceived brightness; strong adjustments can clip colors. The result has the source image's dimensions. Uploaded transparency is not preserved because the image loader's separate alpha mask is not used. This tool applies only color balance; it does not combine the other adjustment tools in this family.
Style Sylva Magic
This is a Textual Inversion trained on one image with DreamArtist, my second publicly released inversion. :)
