# \#ml

**URL:** https://forum.logik.tv/tag/ml/62.md

[Latest](https://forum.logik.tv/latest.md) · [Categories](https://forum.logik.tv/categories.md) · [Tags](https://forum.logik.tv/tags.md)

---

## [VitMatte Matte Refiner in Flame](https://forum.logik.tv/t/vitmatte-matte-refiner-in-flame/14936)

<div class="topic-metadata">

**Author:** [@allklier](https://forum.logik.tv/u/allklier)\
**Replies:** 9\
**Last updated:** [September 21, 2026, 2:52pm UTC](https://forum.logik.tv/t/vitmatte-matte-refiner-in-flame/14936 "2026-09-21T14:52:54Z")

</div>

An often used technique to improve ML mattes is to run them through an edge refiner model. A popular being VitMatte. It needs a trimap of the edge and will then refine the matte further. Here’s an .inf model for VitMat…

---

## [Flame Machine Learning Timewarp, now on Linux and Mac](https://forum.logik.tv/t/flame-machine-learning-timewarp-now-on-linux-and-mac/2038)

<div class="topic-metadata">

**Author:** [@andymilkis](https://forum.logik.tv/u/andymilkis)\
**Replies:** 631\
**Last updated:** [May 13, 2026, 10:23pm UTC](https://forum.logik.tv/t/flame-machine-learning-timewarp-now-on-linux-and-mac/2038 "2026-05-13T22:23:03Z")

</div>

Hi All, Here is the link to the Machine Learning Timewarp tool in Flame that @talosh (Andriy Toloshny) put together. It’s open source, linux only (tested on CentOS 7.4 and 7.6) and provided as-is, so use at your own ris…

---

## [Flame 2026 + AI: real production breakthrough… or still too unstable for high-end workflows?](https://forum.logik.tv/t/flame-2026-ai-real-production-breakthrough-or-still-too-unstable-for-high-end-workflows/14350)

<div class="topic-metadata">

**Author:** [@Alexandre](https://forum.logik.tv/u/Alexandre)\
**Replies:** 13\
**Last updated:** [April 20, 2026, 2:54pm UTC](https://forum.logik.tv/t/flame-2026-ai-real-production-breakthrough-or-still-too-unstable-for-high-end-workflows/14350 "2026-04-20T14:54:04Z")

</div>

I’ve just installed Flame 2026 for a production and started testing the ML tools alongside my usual workflow. I mainly work in advertising, with strong constraints around clean-up, compositing, and especially shot-to-sh…

---

## [Pose Plugin for OFX? Or nah?](https://forum.logik.tv/t/pose-plugin-for-ofx-or-nah/14151)

<div class="topic-metadata">

**Author:** [@eric\_vfxtools](https://forum.logik.tv/u/eric_vfxtools)\
**Replies:** 3\
**Last updated:** [February 23, 2026, 7:15pm UTC](https://forum.logik.tv/t/pose-plugin-for-ofx-or-nah/14151 "2026-02-23T19:15:40Z")

</div>

Is this something you’d want in Flame? If I switch over to MediaPipe, with its open source license, I could port this to OFX. Would that be useful?

---

## [ML tools benchmarks](https://forum.logik.tv/t/ml-tools-benchmarks/13827)

<div class="topic-metadata">

**Author:** [@finnjaeger](https://forum.logik.tv/u/finnjaeger)\
**Replies:** 0\
**Last updated:** [November 26, 2025, 7:51am UTC](https://forum.logik.tv/t/ml-tools-benchmarks/13827 "2025-11-26T07:51:39Z")

</div>

Should we come up with a way to benchmark these AI/ML tools in flame ? in the resolve discord we found that for some tools the M3Ultra was like 2-2.5x as fast as a M2U , namely depthmaps and “Magicmask” . In general co…

---

## [MLMimic](https://forum.logik.tv/t/mlmimic/13754)

<div class="topic-metadata">

**Author:** [@toddnotplaying](https://forum.logik.tv/u/toddnotplaying)\
**Replies:** 4\
**Last updated:** [November 7, 2025, 4:20pm UTC](https://forum.logik.tv/t/mlmimic/13754 "2025-11-07T16:20:57Z")

</div>

uhhh weirdly I think this is my first post after two decades on flame and lurking forever here… so hi everyone long time no see from Todd still out here by the beach in So Cal. Anyone checked out https://www.mlmimic.com…

---

## [9/28 - VES NY Event: Exploring AI in VFX: Innovation, Insights & The State of the Industry](https://forum.logik.tv/t/9-28-ves-ny-event-exploring-ai-in-vfx-innovation-insights-the-state-of-the-industry/13545)

<div class="topic-metadata">

**Author:** [@andymilkis](https://forum.logik.tv/u/andymilkis)\
**Replies:** 4\
**Last updated:** [October 1, 2025, 8:02pm UTC](https://forum.logik.tv/t/9-28-ves-ny-event-exploring-ai-in-vfx-innovation-insights-the-state-of-the-industry/13545 "2025-10-01T20:02:56Z")

</div>

Howdy friends, I’m going to be presenting at this event on Sunday. It’s going to be based on the master class I taught at IBC. If you’re in NY and want to check it out here’s the signup link.

---

## [Would love to do anAI/ML/Comfy/VFX show and tell session at IBC…](https://forum.logik.tv/t/would-love-to-do-anai-ml-comfy-vfx-show-and-tell-session-at-ibc/13331)

<div class="topic-metadata">

**Author:** [@andymilkis](https://forum.logik.tv/u/andymilkis)\
**Replies:** 1\
**Last updated:** [August 5, 2025, 9:20pm UTC](https://forum.logik.tv/t/would-love-to-do-anai-ml-comfy-vfx-show-and-tell-session-at-ibc/13331 "2025-08-05T21:20:16Z")

</div>

Hi friends, I would love to do an AI/ML/Comfy/etc show and tell session at IBC. We get together in the same place and show some examples of how we are using this wild new tech as part of our Flame and VFX pipelines. Ver…

---

## [Flame 2026 BiRefNet - Better Results](https://forum.logik.tv/t/flame-2026-birefnet-better-results/12666)

<div class="topic-metadata">

**Author:** [@allklier](https://forum.logik.tv/u/allklier)\
**Replies:** 6\
**Last updated:** [April 18, 2025, 2:41pm UTC](https://forum.logik.tv/t/flame-2026-birefnet-better-results/12666 "2025-04-18T14:41:57Z")

</div>

With the release of Flame 2026 come some extra Matchboxes that make BiRefNet work with the inference node. Here are some updated tests: The setup requires bracketing the Inference node with PreBRN and PostBRN matchboxe…

---

## [flameSimpleML - Flame Machine Learning Source/Target tool with bespoke training](https://forum.logik.tv/t/flamesimpleml-flame-machine-learning-source-target-tool-with-bespoke-training/9800)

<div class="topic-metadata">

**Author:** [@talosh](https://forum.logik.tv/u/talosh)\
**Replies:** 93\
**Last updated:** [February 1, 2025, 5:27pm UTC](https://forum.logik.tv/t/flamesimpleml-flame-machine-learning-source-target-tool-with-bespoke-training/9800 "2025-02-01T17:27:26Z")

</div>

flameSimpleML - Flame Machine Learning Source/Target tool with bespoke training: Hi guys, I’ve put several scripts together in a package called flameSimpleML: https://github.com/talosh/flameSimpleML/releases/tag/v0.0.1 …

---

## [GPT Flame model - for assist and help](https://forum.logik.tv/t/gpt-flame-model-for-assist-and-help/9479)

<div class="topic-metadata">

**Author:** [@tpo](https://forum.logik.tv/u/tpo)\
**Replies:** 1\
**Last updated:** [November 14, 2023, 11:48am UTC](https://forum.logik.tv/t/gpt-flame-model-for-assist-and-help/9479 "2023-11-14T11:48:06Z")

</div>

Hey guys, ive been talking about this on Discord, putting here… Im current fine-tuning GPT model specific to understand and help FLAME Users, a helper. Im publishing over GPT Builder on OpenAI. https://chat.openai.com…

---

## [ML Face Maps UV Template?](https://forum.logik.tv/t/ml-face-maps-uv-template/8945)

<div class="topic-metadata">

**Author:** [@KuleshovEffect](https://forum.logik.tv/u/KuleshovEffect)\
**Replies:** 0\
**Last updated:** [August 10, 2023, 12:19am UTC](https://forum.logik.tv/t/ml-face-maps-uv-template/8945 "2023-08-10T00:19:27Z")

</div>

ML Face Maps can generate a UV map: If you bring this into Action, you can use it as a UV map and it sticks Ok: But if you wanted to use the UV map to stick some red dots on the face, what’s the UV map using as r…

---

## [Runaway](https://forum.logik.tv/t/runaway/8803)

<div class="topic-metadata">

**Author:** [@Edusanjo](https://forum.logik.tv/u/Edusanjo)\
**Replies:** 3\
**Last updated:** [July 21, 2023, 4:24pm UTC](https://forum.logik.tv/t/runaway/8803 "2023-07-21T16:24:35Z")

</div>

Hi, I just tried Runaway for a difficult clean-up, where the camera is mounted on a drone, pointing to a full-speed jetski, and well… a complex one. We tried the free version to see if it was possible, and we were surpr…

---

## [ChatBot and Python](https://forum.logik.tv/t/chatbot-and-python/8004)

<div class="topic-metadata">

**Author:** [@ytf](https://forum.logik.tv/u/ytf)\
**Replies:** 7\
**Last updated:** [March 23, 2023, 7:39pm UTC](https://forum.logik.tv/t/chatbot-and-python/8004 "2023-03-23T19:39:54Z")

</div>

I wonder if anyone could train a chatbot to write python. Tell it what you want, it writes the code.

---

## [Dreamtextures/ Diffusion models in Flame?](https://forum.logik.tv/t/dreamtextures-diffusion-models-in-flame/7301)

<div class="topic-metadata">

**Author:** [@Ton](https://forum.logik.tv/u/Ton)\
**Replies:** 4\
**Last updated:** [December 16, 2022, 11:53pm UTC](https://forum.logik.tv/t/dreamtextures-diffusion-models-in-flame/7301 "2022-12-16T23:53:48Z")

</div>

As this is all python… would it be far fetched to imagine something like this in Flame?

---

## [ML Full body extraction - Today you are my hero!](https://forum.logik.tv/t/ml-full-body-extraction-today-you-are-my-hero/4378)

<div class="topic-metadata">

**Author:** [@ytf](https://forum.logik.tv/u/ytf)\
**Replies:** 8\
**Last updated:** [November 12, 2021, 7:12pm UTC](https://forum.logik.tv/t/ml-full-body-extraction-today-you-are-my-hero/4378 "2021-11-12T19:12:47Z")

</div>

Doing one of those jobs where everyone is over white with hard edged wipes in between. Of course the set is not perfectly even and almost every shot needs the background replaced with an even flat white. It is taking m…
