NODES: Face Swap, Film Interpolation, Latent Lerp, Int To Number, Bounding Box, Crop, Uncrop, ImageBlur, Denoise, ImageCompare, RGV to HSV, HSV to RGB, Color Correct, Modulo, Deglaze Image, Smart Step, ...

MTB Nodes

Add To Playlist (mtb)


# Animation Builder

Check the
[wiki page](https://github.com/melMass/comfy_mtb/wiki/nodes-animation-builder)
for more info.


- This basic example should help to understand the meaning of
its inputs and outputs thanks to the [debug](nodes-debug) node.

![](https://github.com/melMass/comfy_mtb/assets/7041726/2b5c7e4f-372d-4494-9e73-abb2daa7cb36)

- In this other example Animation Builder is used in combination with
[Batch From History](https://github.com/melMass/comfy_mtb/wiki/nodes-batch-from-history)
to create a zoom-in animation on a static image

![](https://github.com/melMass/comfy_mtb/assets/7041726/77d37da1-0a8e-4519-a493-dfdef7f755ea)

## Inputs

| name | description |
| ---- | :----------:|
| total_frames | The number of frame to queue (this is multiplied by the `loop_count`)|
| scale_float | Convenience input to scale the normalized `current value` (a float between 0 and 1 lerp over the current queue length) |
| loop_count | The number of loops to queue |
| **Reset Button** | resets the internal counters, although the node is though around using its queue button it should still work fine when using the regular queue button of comfy |
| **Queue Button** | Convenience button to run the queues (`total_frames` * `loop_count`) |



Animation Builder (mtb)

# Any to String

Tries to take any input and convert it to a string.


Any To String (mtb)

# Apply Text Template

Very basic string interpolation using [dynamic inputs](web-dynamic-inputs).

The var names are `var_1`, `var_2` etc... They are interpolated like this: `{var_1}`, see the following example for a more concrete idea:

![template_string](https://github.com/melMass/comfy_mtb/assets/7041726/abf8d63e-87f6-426a-adcc-514b85daeb5c)


Apply Text Template (mtb)

Apply Vit Matte (mtb)

Audio Cut (mtb)

Audio Duration (mtb)

Isolate or mute specific speakers using WhisperData

Audio Isolate Speaker (mtb)

Resample audio to a different sample rate.

Audio Resample (mtb)

Sequence audio inputs (dynamic inputs).

    - adding silence_duration between each segment
      can now also be negative to overlap the clips, safely bound
      to the the input length.
    - resample audios to the highest sample rate in the inputs.
    - convert them all to stereo if one of the inputs is.
    

Audio Sequence (mtb)

Stack/Overlay audio inputs (dynamic inputs).

    - pad audios to the longest inputs.
    - resample audios to the highest sample rate in the inputs.
    - convert them all to stereo if one of the inputs is.
    

Audio Stack (mtb)

Audio To Text (mtb)

# Autopan Equilateral

Generate a 360 panning video from an equilateral image.



Auto Pan Equilateral (mtb)

# Batch Transform

Transform a batch of images using a batch of keyframes


Batch2d Transform (mtb)

# Batch Float Assemble

Assembles mutiple batches of floats into a single stream (batch)


Batch Float Assemble (mtb)

# Batch Float Fill

Fills a batch float with a single value until it reaches the target length


Batch Float Fill (mtb)

Fit a list of floats using a source and target range

Batch Float Fit (mtb)

Batch Float Math (mtb)

# Batch Float

Generates a batch of float values with interpolation


Batch Float (mtb)

Normalize the values in the list of floats

Batch Float Normalize (mtb)

# Batch Make 

Simply duplicates the input frame as a batch

Batch Make (mtb)

# Batch Merge

Merges multiple image batches with different frame counts


Batch Merge (mtb)

Sequences multiple image batches one after another

Batch Sequence (mtb)

Sequences multiple image batches with transition effects.

Batch Sequence Plus (mtb)

# Batch Shake

Applies a shaking effect to batches of images.


Batch Shake (mtb)

# Batch Shape

Generates a batch of 2D shapes with optional shading (experimental)


Batch Shape (mtb)

Remap a batch using a time curve (FLOATS)

Batch Time Wrap (mtb)

Batch Transform (mtb)

BBox Force Dimensions (mtb)

# BBox from Mask

From a mask extract the bounding box


Bbox From Mask (mtb)

# BBox

The bounding box (BBOX) custom type used by other nodes


Bbox (mtb)

# Blur

Applies a Gaussian blur to the given image / batch.  
It allows a few more things than the core blur node, namely:
- **Higher values** (core is limited to 30)  
  <img width=650 src="https://github.com/melMass/comfy_mtb/assets/7041726/c0cd8426-7b44-4067-bb98-9a178e715169"/>

- **Individual X & Y blur**  
  <img width=650 src="https://github.com/melMass/comfy_mtb/assets/7041726/f5511f9a-d659-4935-abb5-3242894a0a82"/>

- [`FLOATS`](nodes-batch#batch-floats) support for individual batch frame blur
  <img width=650 src="https://github.com/melMass/comfy_mtb/assets/7041726/8aab18d2-b140-4e04-a4bd-fd294f2e187b"/>




Blur (mtb)

Canny

Various color correction methods using only Torch.

Color Correct GPU (mtb)

# Color Correct

Various color correction methods


Color Correct (mtb)

# Colored Image

Constant color images of a given size can also be used to mask images.

This example uses [VHS Nodes](https://github.com/Kosinkadink/ComfyUI-VideoHelperSuite) for animation preview
![batch_shapes](https://github.com/melMass/comfy_mtb/assets/7041726/a1aa62c5-f8cb-4f90-a14a-f7260a982e5a)

**Workflow**: 
```json
{"last_node_id":41,"last_link_id":65,"nodes":[{"id":13,"type":"Batch Float (mtb)","pos":[-1213,622],"size":[315,154],"flags":{},"order":0,"mode":0,"outputs":[{"name":"FLOATS","type":"FLOATS","links":[30],"shape":3,"slot_index":0}],"properties":{"Node name for S&R":"Batch Float (mtb)"},"widgets_values":["Steps",12,0.001,5,"Quart In/Out"]},{"id":31,"type":"VHS_VideoCombine","pos":[1080,280],"size":[315,314],"flags":{},"order":16,"mode":0,"inputs":[{"name":"images","type":"IMAGE","link":50},{"name":"audio","type":"VHS_AUDIO","link":null},{"name":"batch_manager","type":"VHS_BatchManager","link":null}],"outputs":[{"name":"Filenames","type":"VHS_FILENAMES","links":null,"shape":3}],"properties":{"Node name for S&R":"VHS_VideoCombine"},"widgets_values":{"frame_rate":12,"loop_count":0,"filename_prefix":"AnimateDiff","format":"video/nvenc_hevc-mp4","pix_fmt":"yuv420p","bitrate":10,"megabit":true,"save_metadata":true,"pingpong":false,"save_output":false,"videopreview":{"hidden":false,"paused":false,"params":{"filename":"AnimateDiff_00004.mp4","subfolder":"","type":"temp","format":"video/nvenc_hevc-mp4"}}}},{"id":22,"type":"Batch Float Fill (mtb)","pos":[-884,621],"size":[315,106],"flags":{},"order":4,"mode":0,"inputs":[{"name":"floats","type":"FLOATS","link":30}],"outputs":[{"name":"FLOATS","type":"FLOATS","links":[31],"shape":3,"slot_index":0}],"properties":{"Node name for S&R":"Batch Float Fill (mtb)"},"widgets_values":["tail",0,20]},{"id":30,"type":"ImageBlend","pos":[680,280],"size":[315,102],"flags":{},"order":14,"mode":0,"inputs":[{"name":"image1","type":"IMAGE","link":65},{"name":"image2","type":"IMAGE","link":48}],"outputs":[{"name":"IMAGE","type":"IMAGE","links":[50],"shape":3,"slot_index":0}],"properties":{"Node name for S&R":"ImageBlend"},"widgets_values":[1,"screen"]},{"id":15,"type":"Batch Transform (mtb)","pos":[484,561],"size":[210,194],"flags":{},"order":10,"mode":0,"inputs":[{"name":"image","type":"IMAGE","link":29,"slot_index":0},{"name":"x","type":"FLOATS","link":null},{"name":"y","type":"FLOATS","link":null},{"name":"zoom","type":"FLOATS","link":33},{"name":"angle","type":"FLOATS","link":null},{"name":"shear","type":"FLOATS","link":null}],"outputs":[{"name":"IMAGE","type":"IMAGE","links":[48],"shape":3,"slot_index":0}],"properties":{"Node name for S&R":"Batch Transform (mtb)"},"widgets_values":["edge","#fa96c0"]},{"id":23,"type":"Batch Float Fill (mtb)","pos":[110,623],"size":[315,106],"flags":{},"order":5,"mode":0,"inputs":[{"name":"floats","type":"FLOATS","link":32}],"outputs":[{"name":"FLOATS","type":"FLOATS","links":[33],"shape":3,"slot_index":0}],"properties":{"Node name for S&R":"Batch Float Fill (mtb)"},"widgets_values":["head",0,20]},{"id":21,"type":"Batch Float (mtb)","pos":[-246,623],"size":[315,154],"flags":{},"order":1,"mode":0,"outputs":[{"name":"FLOATS","type":"FLOATS","links":[32],"shape":3,"slot_index":0}],"properties":{"Node name for S&R":"Batch Float (mtb)"},"widgets_values":["Steps",20,0.001,4.5,"Quart In/Out"]},{"id":41,"type":"Reroute","pos":[-91,283],"size":[75,26],"flags":{},"order":12,"mode":0,"inputs":[{"name":"","type":"*","link":64}],"outputs":[{"name":"","type":"IMAGE","links":[65]}],"properties":{"showOutputText":false,"horizontal":false}},{"id":14,"type":"Batch Transform (mtb)","pos":[-517,562],"size":[210,194],"flags":{},"order":9,"mode":0,"inputs":[{"name":"image","type":"IMAGE","link":27,"slot_index":0},{"name":"x","type":"FLOATS","link":null},{"name":"y","type":"FLOATS","link":null},{"name":"zoom","type":"FLOATS","link":31},{"name":"angle","type":"FLOATS","link":null},{"name":"shear","type":"FLOATS","link":null}],"outputs":[{"name":"IMAGE","type":"IMAGE","links":[37,62,64],"shape":3,"slot_index":0}],"properties":{"Node name for S&R":"Batch Transform (mtb)"},"widgets_values":["edge","#fa96c0"]},{"id":35,"type":"LoadImage","pos":[-560,920],"size":[315,314],"flags":{},"order":2,"mode":0,"outputs":[{"name":"IMAGE","type":"IMAGE","links":[56],"shape":3,"slot_index":0},{"name":"MASK","type":"MASK","links":null,"shape":3}],"properties":{"Node name for S&R":"LoadImage"},"widgets_values":["189.png","image"]},{"id":40,"type":"Blur (mtb)","pos":[-217,1000],"size":[315,82],"flags":{},"order":11,"mode":0,"inputs":[{"name":"image","type":"IMAGE","link":62,"slot_index":0}],"outputs":[{"name":"IMAGE","type":"IMAGE","links":[63],"shape":3,"slot_index":0}],"properties":{"Node name for S&R":"Blur (mtb)"},"widgets_values":[50,50]},{"id":33,"type":"ImageToMask","pos":[120,997],"size":[210,58],"flags":{},"order":13,"mode":0,"inputs":[{"name":"image","type":"IMAGE","link":63,"slot_index":0}],"outputs":[{"name":"MASK","type":"MASK","links":[57],"shape":3,"slot_index":0}],"properties":{"Node name for S&R":"ImageToMask"},"widgets_values":["red"]},{"id":37,"type":"Colored Image (mtb)","pos":[350,920],"size":[210,138],"flags":{},"order":15,"mode":0,"inputs":[{"name":"foreground_image","type":"IMAGE","link":56,"slot_index":0},{"name":"foreground_mask","type":"MASK","link":57,"slot_index":1}],"outputs":[{"name":"IMAGE","type":"IMAGE","links":[55],"shape":3,"slot_index":0}],"properties":{"Node name for S&R":"Colored Image (mtb)"},"widgets_values":["#1603fc",512,512]},{"id":36,"type":"VHS_VideoCombine","pos":[584,918],"size":[315,314],"flags":{},"order":17,"mode":0,"inputs":[{"name":"images","type":"IMAGE","link":55,"slot_index":0},{"name":"audio","type":"VHS_AUDIO","link":null},{"name":"batch_manager","type":"VHS_BatchManager","link":null}],"outputs":[{"name":"Filenames","type":"VHS_FILENAMES","links":null,"shape":3}],"properties":{"Node name for S&R":"VHS_VideoCombine"},"widgets_values":{"frame_rate":12,"loop_count":0,"filename_prefix":"AnimateDiff","format":"video/nvenc_hevc-mp4","pix_fmt":"yuv420p","bitrate":10,"megabit":true,"save_metadata":true,"pingpong":false,"save_output":false,"videopreview":{"hidden":false,"paused":false,"params":{"filename":"AnimateDiff_00009.mp4","subfolder":"","type":"temp","format":"video/nvenc_hevc-mp4"}}}},{"id":20,"type":"Mask To Image (mtb)","pos":[231,459],"size":[210,106],"flags":{},"order":8,"mode":0,"inputs":[{"name":"mask","type":"MASK","link":28,"slot_index":0}],"outputs":[{"name":"IMAGE","type":"IMAGE","links":[29],"shape":3,"slot_index":0}],"properties":{"Node name for S&R":"Mask To Image (mtb)"},"widgets_values":["#00ff00","#000000"]},{"id":16,"type":"ImageToMask","pos":[-750,191],"size":[210,58],"flags":{},"order":6,"mode":0,"inputs":[{"name":"image","type":"IMAGE","link":25}],"outputs":[{"name":"MASK","type":"MASK","links":[26,28],"shape":3,"slot_index":0}],"properties":{"Node name for S&R":"ImageToMask"},"widgets_values":["red"]},{"id":19,"type":"Mask To Image (mtb)","pos":[-750,446],"size":[210,106],"flags":{},"order":7,"mode":0,"inputs":[{"name":"mask","type":"MASK","link":26,"slot_index":0}],"outputs":[{"name":"IMAGE","type":"IMAGE","links":[27],"shape":3,"slot_index":0}],"properties":{"Node name for S&R":"Mask To Image (mtb)"},"widgets_values":["#ff0000","#000000"]},{"id":11,"type":"Batch Shape (mtb)","pos":[-1039,192],"size":[210,334],"flags":{},"order":3,"mode":0,"outputs":[{"name":"IMAGE","type":"IMAGE","links":[25],"shape":3,"slot_index":0}],"properties":{"Node name for S&R":"Batch Shape (mtb)"},"widgets_values":[20,"Tube",1024,1024,512,"#ffffff","#000000","#000000",250,0,0]}],"links":[[25,11,0,16,0,"IMAGE"],[26,16,0,19,0,"MASK"],[27,19,0,14,0,"IMAGE"],[28,16,0,20,0,"MASK"],[29,20,0,15,0,"IMAGE"],[30,13,0,22,0,"FLOATS"],[31,22,0,14,3,"FLOATS"],[32,21,0,23,0,"FLOATS"],[33,23,0,15,3,"FLOATS"],[37,14,0,27,0,"IMAGE"],[48,15,0,30,1,"IMAGE"],[50,30,0,31,0,"IMAGE"],[55,37,0,36,0,"IMAGE"],[56,35,0,37,0,"IMAGE"],[57,33,0,37,1,"MASK"],[62,14,0,40,0,"IMAGE"],[63,40,0,33,0,"IMAGE"],[64,14,0,41,0,"*"],[65,41,0,30,0,"IMAGE"]],"groups":[],"config":{},"extra":{"workspace_info":{"id":"lHUoc3eCvYMplIxkGM68o"}},"version":0.4}
```

Colored Image (mtb)

Concat Images (mtb)

Constant (mtb)

Coordinates To String (mtb)

# Crop

Crops an image and an optional mask to a given bounding box

The bounding box can be given as a tuple of (x, y, width, height) or as a BBOX type
The BBOX input takes precedence over the tuple input
    


Crop (mtb)

# Curve

*WIP*
A basic FLOAT_CURVE input node.


Curve (mtb)

Convert a FLOAT_CURVE to a FLOAT or FLOATS

Curve To Float (mtb)

# Debug

This node is basically trying to provide informations about any input types supported by Comfy.
It uses the [dynamic inputs](web-dynamic-inputs) concept used across a few of the mtb nodes.
It specifically handles a few types and fallback to string representation for the others.

![debug](https://github.com/melMass/comfy_mtb/assets/7041726/e20b5379-b5bf-4e2f-8d4c-c79f217d7f8d)


Debug (mtb)

# Deep Bump

This node uses the [deep bump](https://github.com/HugoTini/DeepBump) model (GPLv3).
The 3 inference modes (color -> normals, normals -> curvature, normals -> depth) are all baked into a single node with a dropdown to select the operation.
Some inputs are only used in some context, UX could be better.
The inputs are self explanatory, but you should probably experiment a bit with it since inference is quite fast. One thing to be sure is to tick `normals_to_height_seamless` when the input is seamless, see below for more infos.

This example is available in the [base examples list](https://github.com/melMass/comfy_mtb/wiki/Examples). In the example we also use the [Model Patch Seamless](nodes-model-patch-seamless) node in order to have non repeating, tileable textures
| workflow | This is the output textures from the workflow applied to a tessellated mesh in blender | 
| - | - |
|![](https://user-images.githubusercontent.com/7041726/272970715-7e4477f6-8e18-4839-9864-83d07d6690a1.png)| ![](https://user-images.githubusercontent.com/7041726/272970506-9db516b5-45d2-4389-b904-b3a94660f24c.png) |

Deep Bump (mtb)

# Export with FFmpeg

Quite crude node, [VHS](https://github.com/Kosinkadink/ComfyUI-VideoHelperSuite) is recommended now


Export With Ffmpeg (mtb)

Extract 2D points from a batch of images based on a threshold.

Extract Coordinates From Image (mtb)

# Face Swap
Face swap using deepinsight/insightface models


Face Swap (mtb)

# Film Interpolation

> **Warning**
> This is part of the `optional` nodes, if you are on linux or mac you can skip this warning, on windows I would recommend looking into alternative like [ComfyUI-Frame-Interpolation](https://github.com/Fannovel16/ComfyUI-Frame-Interpolation) 

## Load Film Model
This node loads and cache a given FILM model


## FILM interpolation

Wrapper of the [original implementation in tensorflow](https://github.com/google-research/frame-interpolation). You should not use tensorflow CPU as this operation is really slow on CPU and quite fast on GPU

![save_gif](https://github.com/melMass/comfy_mtb/assets/7041726/a74def82-7c0f-4907-9a52-2d28ddd6306c)






Film Interpolation (mtb)

# Filter Z
Filters an image based on a depth map


Filter Z (mtb)

Fit the input float using a source and target range

Fit Number (mtb)

# `FLOATS` to `FLOAT`

AD, IPA, Fitz etc have commonly choose to mistype float lists as FLOAT.

This is just a hack to be compatible with these
    


Floats To Float (mtb)

Conversion utility for compatibility with frame interpolation.

Floats To Ints (mtb)

Conversion utility for compatibility with other extensions (AD, IPA, Fitz are using FLOAT to represent list of floats.)

Float To Floats (mtb)

Node addon for the WAS Suite. Converts a "comfy" FLOAT to a NUMBER.

Float To Number (mtb)

Generate Trimap (mtb)

# Get Batch from History

This experimental node does something really simple, it reads the outputs from the history endpoint of Comfy.
Outputs gets populated by... output nodes. There are various ones but  for instance in core comfy, `Save Image` and `Preview Image` are output nodes.
I advice to start simple and have workflows that only generates one output per queue run. Of course once you master it you can use multiple outputs as output order is kept (as long as all outputs are ran).

Another basic use case of batch from history that you can see in the 4th [example](Examples), the fake deforum effect, basically this flow allows you to **feedback** an image using the history.

A classic example when showing the feedback concept is the poor man's [grey scott diffusion model](https://groups.csail.mit.edu/mac/projects/amorphous/GrayScott/) i.e the "creative" derivative using only a gaussian blur and a sharp at each fed steps. 

 
Here is an example workflow of just that.

this is the output:  
<p align=center>
<img width=250 src="https://github.com/melMass/comfy_mtb/assets/7041726/162fb62e-96d4-4843-8902-19d59c536469"/>
</p>

and the workflow:
<p align=center>  
<img width=1000 src="https://github.com/melMass/comfy_mtb/assets/7041726/11257918-86ff-4ddd-8d08-f2a59f7f45a2"/>
</p>

<details><summary>expand here to copy paste this workflow</summary>

```json
{"last_node_id":17,"last_link_id":27,"nodes":[{"id":9,"type":"Get Batch From History (mtb)","pos":[181,706],"size":[315,130],"flags":{},"order":5,"mode":0,"inputs":[{"name":"passthrough_image","type":"IMAGE","link":null,"slot_index":0,"shape":7},{"name":"enable","type":"BOOLEAN","link":6,"widget":{"name":"enable"}}],"outputs":[{"name":"images","type":"IMAGE","links":[23],"slot_index":0,"shape":3}],"properties":{"Node name for S&R":"Get Batch From History (mtb)"},"widgets_values":[true,44,0,969]},{"id":13,"type":"Sharpen (mtb)","pos":[996,276],"size":[315,130],"flags":{},"order":9,"mode":0,"inputs":[{"name":"image","type":"IMAGE","link":14}],"outputs":[{"name":"IMAGE","type":"IMAGE","links":[22],"slot_index":0,"shape":3}],"properties":{"Node name for S&R":"Sharpen (mtb)"},"widgets_values":[31,2,1,1]},{"id":15,"type":"VHS_VideoCombine","pos":[528,708],"size":[276,580],"flags":{},"order":7,"mode":0,"inputs":[{"name":"images","type":"IMAGE","link":23},{"name":"audio","type":"AUDIO","link":null,"shape":7},{"name":"meta_batch","type":"VHS_BatchManager","link":null,"shape":7},{"name":"vae","type":"VAE","link":null,"shape":7}],"outputs":[{"name":"Filenames","type":"VHS_FILENAMES","links":null}],"properties":{"Node name for S&R":"VHS_VideoCombine"},"widgets_values":{"frame_rate":20,"loop_count":0,"filename_prefix":"MTB_BatchFromHistory","format":"video/h264-mp4","pix_fmt":"yuv420p","crf":19,"save_metadata":true,"pingpong":false,"save_output":true,"videopreview":{"hidden":false,"paused":false,"params":{"filename":"AnimateDiff_01089.mp4","subfolder":"","type":"output","format":"video/h264-mp4","frame_rate":20},"muted":false}}},{"id":16,"type":"LoadImage","pos":[-168,-216],"size":[315,314],"flags":{},"order":0,"mode":4,"inputs":[],"outputs":[{"name":"IMAGE","type":"IMAGE","links":[25],"slot_index":0},{"name":"MASK","type":"MASK","links":null}],"properties":{"Node name for S&R":"LoadImage"},"widgets_values":["example.png","image"]},{"id":17,"type":"ImageScale","pos":[192,-36],"size":[315,130],"flags":{},"order":3,"mode":4,"inputs":[{"name":"image","type":"IMAGE","link":25}],"outputs":[{"name":"IMAGE","type":"IMAGE","links":[],"slot_index":0}],"properties":{"Node name for S&R":"ImageScale"},"widgets_values":["nearest-exact",512,512,"disabled"]},{"id":10,"type":"Blur (mtb)","pos":[672,276],"size":[315,122],"flags":{},"order":8,"mode":0,"inputs":[{"name":"image","type":"IMAGE","link":24},{"name":"sigmasX","type":"FLOATS","link":null,"shape":7},{"name":"sigmasY","type":"FLOATS","link":null,"shape":7}],"outputs":[{"name":"IMAGE","type":"IMAGE","links":[14],"slot_index":0,"shape":3}],"properties":{"Node name for S&R":"Blur (mtb)"},"widgets_values":[6,6]},{"id":3,"type":"Get Batch From History (mtb)","pos":[336,276],"size":[315,130],"flags":{},"order":6,"mode":0,"inputs":[{"name":"passthrough_image","type":"IMAGE","link":27,"slot_index":0,"shape":7},{"name":"enable","type":"BOOLEAN","link":16,"widget":{"name":"enable"}}],"outputs":[{"name":"images","type":"IMAGE","links":[24],"slot_index":0,"shape":3}],"properties":{"Node name for S&R":"Get Batch From History (mtb)"},"widgets_values":[false,1,0,969]},{"id":6,"type":"Int To Bool (mtb)","pos":[324,444],"size":[210,42.27488708496094],"flags":{},"order":4,"mode":0,"inputs":[{"name":"int","type":"INT","link":4,"widget":{"name":"int"}}],"outputs":[{"name":"BOOLEAN","type":"BOOLEAN","links":[16],"slot_index":0,"shape":3}],"properties":{"Node name for S&R":"Int To Bool (mtb)"},"widgets_values":[0]},{"id":2,"type":"PreviewImage","pos":[972,456],"size":[360,348],"flags":{},"order":10,"mode":0,"inputs":[{"name":"images","type":"IMAGE","link":22}],"outputs":[],"properties":{"Node name for S&R":"PreviewImage"},"widgets_values":[]},{"id":1,"type":"Batch Shape (mtb)","pos":[-120,192],"size":[210,334],"flags":{},"order":1,"mode":0,"inputs":[],"outputs":[{"name":"IMAGE","type":"IMAGE","links":[27],"slot_index":0,"shape":3}],"properties":{"Node name for S&R":"Batch Shape (mtb)"},"widgets_values":[1,"Diamond",512,512,229,"#ffffff","#000000","#000000",0,0,0]},{"id":4,"type":"Animation Builder (mtb)","pos":[-110,596],"size":[210,318],"flags":{},"order":2,"mode":0,"inputs":[],"outputs":[{"name":"frame","type":"INT","links":[4],"slot_index":0,"shape":3},{"name":"0-1 (scaled)","type":"FLOAT","links":null,"shape":3},{"name":"count","type":"INT","links":null,"shape":3},{"name":"loop_ended","type":"BOOLEAN","links":[6],"slot_index":3,"shape":3}],"properties":{"Node name for S&R":"Animation Builder (mtb)"},"widgets_values":[45,1,1,0,0,null,null,"reset","queue"]}],"links":[[4,4,0,6,0,"INT"],[6,4,3,9,1,"BOOLEAN"],[14,10,0,13,0,"IMAGE"],[16,6,0,3,1,"BOOLEAN"],[22,13,0,2,0,"IMAGE"],[23,9,0,15,0,"IMAGE"],[24,3,0,10,0,"IMAGE"],[25,16,0,17,0,"IMAGE"],[27,1,0,3,0,"IMAGE"]],"groups":[],"config":{},"extra":{"ds":{"scale":0.6727499949325705,"offset":[682.4627866508608,403.88645096355583]},"ue_links":[]},"version":0.4}
```

</details>

The blue bordered node is the one doing the feedback, on first frame (frame == 0 converted to bool is false) the passthrough image will be used, this example uses the [Batch Shape](nodes-batch-shape) node, only on the first queue item, then the previous queue item is fed to each subsequent queue item.
The orange bordered one is fetching all the frames we queued once done to assemble the GIF. All this happens in "one click" thanks to [Animation Builder](nodes-animation-builder)


## Inputs
|name|description|
|-|-|
|passthrough_image | This is the image that gets sent out when `enable` is set to false, useful for the init first image in the fake deforum [example](Examples) for instance ([04-animation_builder-deforum.json](https://github.com/melMass/comfy_mtb/blob/main/examples/04-animation_builder-deforum.json)) |
|enable | This makes the node not fetch the history. For instance when you just initiated the server the history is empty, see [Animation Builder](nodes-animation-builder) for practical examples |
|count | the number of frames to fetch from the history |
| **Reset Button** | resets the internal counters, although the node is though around using its queue button it should still work fine when using the regular queue button of comfy |
| **Queue Button** | Convenience button to run the queues (`total_frames` * `loop_count`) |

Get Batch From History (mtb)


    # Image Batch To Sublist 🔄

    Splits a large batched tensor into smaller sub-batches for memory-efficient processing.
    

Image Batch To Sublist (mtb)

# Image Compare
Compare two images and return a difference image


Image Compare (mtb)


**Encodes the input with h264 compression using a configurable CRF**.

> [!IMPORTANT]
> This node is not really needed with the latest version of LTXVideo.

> [!NOTE]
> This was recommended by the creators of LTX over banodoco's discord.

*Orginal code from [mix](https://github.com/XmYx)*

Image H264 Compression (mtb)

# Image Premultiply

Premultiply image with mask


Image Premultiply (mtb)

# Image Remove Background

Removes the background from the input using Rembg.


Image Remove Background Rembg (mtb)

# Image Resize Factor

Extracted mostly from WAS Node Suite, with a few edits (most notably multiple image support) and less features.


Image Resize Factor (mtb)

# Image Tile Offset

Mimics an old photoshop technique to check for seamless textures


Image Tile Offset (mtb)

Interpolate Clip Sequential (mtb)

Interpolate Condition (mtb)

# Int to Bool

Basic int to bool conversion, >= 1 is true


Int To Bool (mtb)

Node addon for the WAS Suite. Converts a "comfy" INT to a NUMBER.

Int To Number (mtb)

# Latent Lerp
Linear interpolation (blend) between two latent vectors


Latent Lerp (mtb)

# Load Face Analysis

Loads a face analysis model


Load Face Analysis Model (mtb)

# Load Face Enhance Model

Loads a GFPGan or RestoreFormer model for face enhancement.


Load Face Enhance Model (mtb)

# Load Face Swap Model

Loads a faceswap model


Load Face Swap Model (mtb)

Load Film Model (mtb)

# Load image from URL

Load an image from the given URL


Load Image From Url (mtb)

# Load Image Sequence

Load an image sequence from a folder. The current frame is used to determine which image to load.  
The UX need improvements but you can use it as follow:  

- If current_frame is -1, it will load all the frames matching the pattern.
- If the path contains a `*` it will glob the paths using it.
- If range is provided (for instance `0-10` to load frame 0 to 10) current_frame is ignored.



Load Image Sequence (mtb)

Load Vit Matte Model (mtb)

Load Whisper (mtb)

# Mask to Image

Converts a mask (alpha) to an RGB image with a color and background


Mask To Image (mtb)

# Match Dimensions

Match images dimensions along the given axis, preserving aspect ratio.


Match Dimensions (mtb)

evaluate a simple math expression string, only supports literal_eval

Math Expression (mtb)

# Model Patch Seamless

This uses this [hack](https://gitlab.com/-/snippets/2395088) to generate seamless image right at the inference stage.
Results might vary depending on the model and prompt.

Here is a few output from an extended version of the available [example](https://github.com/melMass/comfy_mtb/wiki/Examples). The main difference is that I use an upscale step before running [deep bump](nodes-deep-bump).

|albedo|
|-|
|<img width=400 src="https://github.com/melMass/comfy_mtb/assets/7041726/3984907b-617e-49ff-8100-92c91af4459e"/>|
|<img width=400 src="https://github.com/melMass/comfy_mtb/assets/7041726/1194b2b8-eca4-4f75-8a1c-4d8074c08ea6"/>|
|<img width=400 src="https://github.com/melMass/comfy_mtb/assets/7041726/11026b5d-500b-4cfe-8cdc-682cc995dfb1"/>|
|<img width=400 src="https://github.com/melMass/comfy_mtb/assets/7041726/1f5d1671-5208-47ca-b625-478b09eed969"/>|
|<img width=400 src="https://github.com/melMass/comfy_mtb/assets/7041726/cdf3463f-c66c-472c-8ac3-e80af1901852"/>|
|<img width=400 src="https://github.com/melMass/comfy_mtb/assets/7041726/151b44ca-26e2-49d0-91be-cda938c0577a"/>|
|<img width=400 src="https://github.com/melMass/comfy_mtb/assets/7041726/b4280a65-644f-45d8-9fe8-b6148bf66b3c"/>|
|<img width=400 src="https://github.com/melMass/comfy_mtb/assets/7041726/cbfacbbc-2ae2-4cd2-84bc-8849c2115b51"/>|
|<img width=400 src="https://github.com/melMass/comfy_mtb/assets/7041726/8c6eea98-ae90-48d3-9051-de542102166e"/>|

Model Patch Seamless (mtb)

# Model Pruner

Basic output node to prune/downsample a model.

If `save_folder` is a relative path, it will be relative to comfy's output directory?

**This is a bit experimental for now**


Model Pruner (mtb)

# Pick From Batch
Pick a specific number of images from a batch either from the head or the tail.
    


Pick From Batch (mtb)

# Plot `FLOATS`
Plot FLOATS using matplotlib, each of them are drawn in a different
color.


Plot Batch Float (mtb)

Postshot Export (mtb)

# Postshot Train

https://github.com/user-attachments/assets/fcaf4163-28c7-4768-a785-bc794fca3ec0


Wrapper around the [Potshot](https://www.jawset.com/) CLI. 
You must first run the GUI at least once and login.
If you use a custom install location you can specify it in **Settings > MTB**



## Radiance Field Profile

Postshot supports two different models to create radiance fields:
- Gaussian Splatting (Splat)
- Neural Radiance Fields (NeRF).

### Splat MCMC

Both Splat profiles allow for very fast rendering and quickly reconstruct fine detail in well-covered regions of the scene.

The Splat MCMC profile is currently the recommended profile for most scenes. It allows limiting the number of Splat primitives and thereby the amount of memory and disk space the resulting model requires.

### Splat ADC

The Splat ADC profile is very similar to the Splat MCMC profile, but differs in the way it produces detail in the scene during training. You can control the amount of detail it creates during training through the Splat Density parameter.

### NeRF models

When using the NeRF model, the maximum accuracy has to be specified before the training can begin. Postshot currently provides five sizes (S, M, L, XL, XXL) for NeRF models. NeRFs are much slower to render than Splats.

Here is an intuition for how 'large' the NeRF profile options are:
- **S** is for toy-like testing.
- **M** is a significant step up, such that real scenes can be reasonably captured with low memory requirements.
- **L** is the recommended default if you want to produce good image quality.
- **XL** and **XXL** are for pushing toward fine detail in the scene center or for large scenes.


Postshot Train (mtb)

Process Whisper chunks with speaker diarization using either pyannote or NeMo.

Process Whisper Diarization (mtb)

Process Whisper output into timestamped chunks.

Process Whisper Output (mtb)

Qr Code (mtb)

Read Playlist (mtb)

# Restore Face
Uses GFPGan to restore faces


Restore Face (mtb)

# Save GIF
Save the images from the batch as a GIF


Save Gif (mtb)

# Save Image Grid

Save all the images in the input batch as a grid of images.


Save Image Grid (mtb)

Saves the input images to your ComfyUI output directory.
    This behaves exactly like the native SaveImage node but isn't an output node.
    The reason I made this is to allow 'inlining' image save in loops for instance,
    using the native node there wouldn't run for each iteration of the loop.

Save Image (mtb)

# Save Image Sequence

Save an image sequence to a folder. The current frame is used to determine which image to save.
This is merely a wrapper around the `save_images` function with formatting for the output folder and filename.
    


Save Image Sequence (mtb)

# Save Tensors
Save torch tensors (image, mask or latent) to disk.
useful to debug things outside comfy like in notebooks.
    


Save Tensors (mtb)

# Sharpen
Sharpens an image using a Gaussian kernel.


Sharpen (mtb)

# Smart Step

Utils to control the steps start/stop of the KAdvancedSampler in percentage


Smart Step (mtb)

Split Bbox (mtb)

# Stack Images

A simple way to stack images either horizontally or vertically. Stack image uses [dynamic inputs](web-dynamic-inputs).

It outputs RGBA tensors and supports RGB or RGBA as input (normalized to RGBA internally). If the image dimensions don't match they must at least match:
- in `width` when stacking vertically.
- in `height` when stacking horizontally.

Here is an example workflow using [Text To Image](nodes-text-to-image) (the text was generated using [Nous Hermes 2 Vision](https://huggingface.co/billborkowski/llava-NousResearch_Nous-Hermes-2-Vision-GGUF) thanks to the great [ComfyUI_VLM_nodes](https://github.com/gokayfem/ComfyUI_VLM_nodes) extension. For simplicity's sake, the workflow doesn't contain external nodes:

![stack_images](https://github.com/melMass/comfy_mtb/assets/7041726/a0c03621-3377-46cf-a6b4-f47e70c7d11f)


Stack Images (mtb)

# String Replace
Basic string replacement.


ComfyUI Extension: MTB Nodes

Custom Nodes (110)

README

MTB Nodes