Utility nodes for ComfyUI, in one small pack.
| Registration key | Display name | What it does |
|---|---|---|
BC_AutoBypass |
Auto Bypass | Bypasses its targets automatically when a watched source is empty |
BC_LogicBoolean |
Logic Boolean | 0–1 float → BOOLEAN / NUMBER / INT / FLOAT |
BC_IsMaskEmpty |
Is Mask Empty | MASK → BOOLEAN |
BC_MaskFillHoles |
Mask Fill Holes | Fills enclosed holes in a mask |
BC_MaskGrow |
MaskGrow | Grows / shrinks a mask, then blurs it |
BC_DrawMaskOnImage |
Draw Mask On Image | Paints a colour (with opacity) through a mask onto an image batch |
BC_BlockifyMask |
Blockify Mask | A mask as the blocks of its bounding box that hold any of it |
BC_RepeatMaskBatch |
Repeat Mask Batch | A mask batch repeated amount times |
BC_ImageScaleByAspectRatio |
Image Scale By Aspect Ratio | Scales an image / mask to an aspect ratio and side length; MASK output is never None |
BC_ImageResize |
Image Resize | Resizes an image batch (and mask) to a width and height: stretch, keep the proportion, pad or crop |
BC_JoinImageLists |
Join Image Lists | Concatenates image lists, unlimited inputs |
BC_MathExpression |
Math Expression | Arithmetic over a, b, c without eval() |
BC_PromptList |
Prompt List | One prompt per line, as a list |
BC_AspectPromptList |
Aspect Prompt List | Prompts grouped under [WxH] headers, as prompt / width / height lists |
BC_AnySwitch |
Any Switch | First connected non-None input, any type, unlimited inputs |
BC_SelectSwitch |
Select Switch | Input of the selected named option, any type; only the selected branch runs |
BC_Seed |
Seed | Seed widget; -1 draws a new random seed on every run |
BC_ShowText |
Show Text | Shows incoming text on the node, passes it on |
BC_QwenLM |
Qwen LM | Runs a Qwen chat model (Qwen3.5-4B / 9B, Qwen3.8-27B, Qwen3-VL-8B, and abliterated builds of Qwen3.5-9B and Qwen3-VL-8B) through ComfyUI core: system / user / prefill and images in, text and thinking out, with the official chat templates and sampling defaults and one LoRA; the model file is found in models/Qwen-LM or models/text_encoders, or downloaded on first use |
BC_LMConfig |
LM Config | Sampling settings for an LM node: only the fields you change or link replace the model's official defaults |
BC_LMLoader |
LM Loader | Picks the chat model a wrapper node runs (Captioner, Prompter, PreFlight Observe): family, model, precision, one LoRA and its strength, keep loaded; loads nothing itself, and its lists follow the selected family and model |
BC_Captioner |
Captioner | One training caption per image frame with the model of a connected LM Loader: a built-in tags or prose prompt, or the image alone (none) for your own image-to-caption LoRA, optionally with an instruction or an LM Config; outputs the frames and their captions as lists, ready for Save Image With Caption |
BC_Prompter |
Prompter | One text-to-image prompt from a short brief and / or a reference image with the model of a connected LM Loader: a built-in prompt that writes the person's appearance (include identity) or leaves it to a character LoRA (exclude identity), or none for your own prompter LoRA; text out, and with a CLIP connected the positive conditioning as CLIP Text Encode encodes it |
BC_PreFlightObserve |
PreFlight Observe | Runs the LM Loader's chat model (default Qwen3.5-9B INT8 ConvRot) as a pure observation sensor: an image or frame batch → a strict observations JSON (garment, exposure, framing, pose, in-image text, …); greedy, so the same input gives the same output; an error comes back as {"error": ...} |
BC_PreFlightReport |
PreFlight Report | The rules engine: observations + caption → per-platform verdict ranges (Instagram / TikTok / X: OK, RISK = reach demotion, BLOCK = removal), a summary and the prediction logged to the feedback store |
BC_PreFlightOutcome |
PreFlight Outcome | Records what a platform actually did to a published post (clean / demoted / removed); its record list shows each new prediction at once |
BC_PreFlightCalibrate |
PreFlight Calibrate | The feedback store as two calibration tables (per platform, per rule), as text for Show Text; runs again only when the store changed |
BC_ImageComparer |
Image Comparer | Two images on the node, compared with a sliding divider |
BC_PowerLoraLoader |
Power Lora Loader | MODEL + any number of LoRA rows → MODEL, no CLIP; each LoRA with the key fix of Lora Loader (Key Fix) |
BC_LoraLoaderKeyFix |
Lora Loader (Key Fix) | MODEL + one LoRA → MODEL, with the keys ComfyUI's own loader leaves out renamed first (lightx2v .diff_m, PEFT keys without the diffusion_model. prefix); the console says how many were renamed and how many match no module |
BC_PatchSageAttention |
Patch Sage Attention | MODEL → a clone whose attention runs one SageAttention kernel, for that model only (no global --use-sage-attention); auto takes the most accurate kernel, qk_quant its Q·K quantization (per_thread / per_warp), and a call the kernel cannot take, a kernel error or a NaN output runs the original attention; the console counts both per run |
BC_AnythingEverywhere |
Anything Everywhere | Feeds unconnected inputs of a type at prompt time |
BC_FastGroupsBypasser |
Fast Groups Bypasser | One bypass toggle per group |
BC_BiRefNetRemoveBackground |
BiRefNet Remove Background | Background removal with BiRefNet, plain torch |
BC_DepthAnythingV2 |
Depth Anything | Depth map with Depth Anything V2 Small or Depth Anything 3 (Small, Base, Mono-Large, Metric-Large; all Apache-2.0), near = white, at a ControlNet preprocessor's size or exactly width x height |
BC_SeedVR2FramingDownscale |
SeedVR2 Framing Downscale | SeedVR2 Resize's downscale_factor from the face size: the tallest SAM 3 face box picks close-up, medium or far, each with its factor; the measured face fraction comes out too |
BC_SeedVR2Resize |
SeedVR2 Resize | Original image → the padded frame SeedVR2 encodes (lanczos downscale, shortest-edge antialiased bicubic, pad 16, 4n+1 frames) plus the colour reference |
BC_SeedVR2VAEEncode |
SeedVR2 VAE Encode | SeedVR2 VAE encode with the frames streamed from RAM slice by slice, so VRAM does not grow with the frame count |
BC_SeedVR2ChunkSize |
SeedVR2 Chunk Size | The longest 4n+1 chunk the SeedVR2 sampler fits on this card, from the latent's size and the card's total memory; wire to Split SeedVR2 Latent's manual frames_per_chunk |
BC_SeedVR2VAEDecode |
SeedVR2 VAE Decode | SeedVR2 VAE decode with every decoded slice streamed to RAM, so VRAM does not grow with the frame count |
BC_SeedVR2PostProcess |
SeedVR2 PostProcess | Post-Process SeedVR2 Output one frame at a time into one float16 output, no full-size temporaries |
BC_SeedVR2PreprocessCompact |
SeedVR2 Preprocess (Compact) | SeedVR2 Resize and VAE Encode in one node: outputs only the latent and the plan PostProcess (Compact) reads, no colour reference kept |
BC_SeedVR2PostProcessCompact |
SeedVR2 PostProcess (Compact) | SeedVR2 VAE Decode and PostProcess in one node: decodes only the kept frames and corrects them in place against a reference rebuilt per chunk from the original frames |
BC_AutoModelDownloader |
Auto Model Downloader | Lists a workflow's models and fetches the missing ones into models/ |
BC_PostFxApply |
PostFx Apply | Applies a postfx film-emulation look (theme + condition + strength) to an image batch, optional look override and mask |
BC_PostFxTheme |
PostFx Theme | Built-in postfx theme → POSTFX_LOOK, to start a chain from a named look |
BC_PostFxCustomLook |
PostFx Custom Look | Builds a look from common controls, or overrides them on top of an incoming look |
BC_PostFxLut |
PostFx LUT | 3D .cube LUT as a look, standalone or layered on a theme |
BC_PostFxSignatureSheet |
PostFx Signature Sheet | Labeled contact sheet of every theme in a category applied to one image |
BC_CaptionAudit |
Caption Audit | Runs caption-audit over a LoRA caption folder and draws the report card on the node |
BC_SocialMediaExport |
Social Media Export | One platform-ready derivative per ticked platform, minimum crop, spec-driven |
BC_ImageQualityGate |
Image Quality Gate | Blur / sharpness / noise / clipping / entropy → PASS / SO-SO / FAIL badge, verdict and scores |
BC_SaveImage |
Save Image | Saves images with folder / file names built from prompt widget values, any Pillow format, prompt + workflow embedded; preview only in the gallery, never under the node |
BC_SaveImageWithCaption |
Save Image With Caption | Saves images as PNG and, with a caption connected, the caption next to each image under the same name, for training datasets |
BC_SkinTexture |
Skin Texture | Micro-texture on skin inside a SAM 3 mask: boosts the image's own detail and multiplies in a synthetic pore field, in linear light |
BC_FrequencyMerge |
Frequency Merge | The low frequencies (structure, colour) of one image + the high frequencies (texture, fine detail) of another of the same size, split by a Gaussian |
| — | Align | Align / distribute buttons in the selection toolbox |
| — | Process Monitor | Not a node: live RAM / VRAM bars, an estimate before a run, per-node measurement, the reason a killed run died, a full clear back to the startup memory without a restart, and a drop stale outputs setting (off) (see Process Monitor) |
Registration keys are BCNodes' own, so the packages above can be installed side by side without a clash. Type BCNodes in the node library to see them all; in the menu they sit in these groups:
| Category | Nodes |
|---|---|
BCNodes/logic |
Logic Boolean, Math Expression, Any Switch, Select Switch, Seed |
BCNodes/mask |
Mask Fill Holes, MaskGrow, Draw Mask On Image, Blockify Mask, Repeat Mask Batch, Is Mask Empty, BiRefNet Remove Background |
BCNodes/image |
Image Scale By Aspect Ratio, Image Resize, Join Image Lists, Depth Anything, Social Media Export, Save Image, Save Image With Caption, Skin Texture, Frequency Merge |
BCNodes/postfx |
PostFx Apply, Theme, Custom Look, LUT, Signature Sheet |
BCNodes/analysis |
Image Quality Gate, Caption Audit |
BCNodes/text |
Prompt List, Show Text |
BCNodes/lm |
Qwen LM, LM Config, LM Loader, Captioner, Prompter |
BCNodes/preflight |
PreFlight Observe, Report, Outcome, Calibrate |
BCNodes/loaders |
Power Lora Loader, Lora Loader (Key Fix), Auto Model Downloader |
BCNodes/model |
Patch Sage Attention |
BCNodes/seedvr2 |
SeedVR2 Framing Downscale, Resize, VAE Encode, Chunk Size, VAE Decode, PostProcess; Compact: Preprocess (Compact), PostProcess (Compact) |
BCNodes/workflow |
Image Comparer, Anything Everywhere, Fast Groups Bypasser, Auto Bypass |
cd ComfyUI/custom_nodes
git clone https://github.com/0xBeycan/ComfyUI-BCNodes
pip install -r ComfyUI-BCNodes/requirements.txt
Restart ComfyUI. requirements.txt holds opencv-python (the morphology of MaskGrow and the hole filling of Mask Fill Holes) and the two packages that back the PostFx and Caption Audit nodes, postfx and caption-audit; everything else ships with ComfyUI. Without postfx and caption-audit the pack still loads and only those nodes are absent. Patch Sage Attention needs SageAttention 2.2.x, which is not in requirements.txt (see Model); without it the node leaves the model on ComfyUI's attention. If this repository is still present under its old name ComfyUI-AutoBypass, delete that folder — its AutoBypass node is this pack's BC_AutoBypass.
The Align buttons are not a node; they appear in the toolbox above a multi-selection.
- Logic — booleans, arithmetic, switches and the seed
- Mask — mask checks, fill / grow / draw / blockify / repeat, BiRefNet background removal
- Image — scaling and resizing, image lists, depth maps, social media export, saving, skin texture, frequency merge
- PostFx — film-emulation looks: themes, custom looks, LUTs, signature sheets
- Analysis — the caption set audit and the image quality gate
- Text — prompt lists and showing text
- LM — chat models through ComfyUI core: Qwen LM, LM Config, LM Loader, Captioner, Prompter, the models and their files, sampling, LoRAs,
models.yaml - PreFlight — how Instagram, TikTok and X would likely treat an image or video before it is published: the observation sensor, the rules engine, reading a verdict, the feedback loop and calibration
- Loaders — LoRA loading (with the key fix for LoRAs core loads only in part) and the model downloader
- Model — SageAttention for one model in one workflow: the modes,
qk_quant, long sequences, the calls that run the original attention,nan_guard, the console, install, an A/B test - SeedVR2 — the downscale factor from the framing, the SeedVR2 input stage, streaming VAE encode / decode, the chunk size and the VAE tile for the card, post-processing
- Workflow — image comparer, Anything Everywhere, group bypass, Auto Bypass, the Align buttons
- Unused outputs — unconnected whole-batch outputs come out empty
- Measurements — time, RAM and output size of the mask nodes and Image Resize against the nodes they replace
- Process Monitor — RAM / VRAM live, an estimate before a run, per-node measurement, the report of a killed run, the full clear
- Development — the repository layout, the tests and the import rule
models/birefnet/arch/— the BiRefNet architecture and its Swin v1 backbone, MIT, notices inmodels/birefnet/arch/LICENSE.models/depth_anything_v2/arch/— the Depth Anything V2 DPT head and its DINOv2 ViT-S backbone, Apache-2.0, notices and license text inmodels/depth_anything_v2/arch/LICENSE.
Everything else in this pack is BCNodes' own code.
MIT. The code in models/birefnet/arch/ keeps its own MIT notice, in models/birefnet/arch/LICENSE; the code in models/depth_anything_v2/arch/ keeps its Apache-2.0 license, in models/depth_anything_v2/arch/LICENSE.