exportAnything

ComfyUI power tools & LTX Director nodes • Chrome extension developer • Secretly building something huge for 2027

[exportAnything @exportAnything]

Single stage #LTX reframe workflow in @ComfyUI at 297 seconds using a custom Reframe node. The output video quality isn't as good as it can be, but an additional second refinement stage may fix this.

#ComfyUI #reframe #aivideos

[Robin Huang @robinjhuang]

And it shipped right. Example files, pre-toggled settings, real VRAM numbers from actual testing.

Your first run matches his demo exactly.

Repo: https://t.co/wiDAUO5L4L

[exportAnything @exportAnything]

@robinjhuang @robinjhuang Thanks a ton for the shoutout, Robin! Super happy you see the value in it. 🙌

[Robin Huang @robinjhuang]

In AI video gen you spend more time describing your reference than actually generating.

Someone just fixed that.

[jtydhr88 @jtydhr88]

ComfyUI Native 3D Camera Node.

Adding intuitive controls to the camera node.
Actually, this node wasn’t added by me, it was put in by KJ (yes, the KJ everyone knows) while he was working on TripoSplat.

That said, KJ only built the backend and a basic numeric panel. Honestly, after trying it out, I felt that no one is really going to adjust the camera’s position and view angle through such an unintuitive method.

So a visual 3D camera adjustment became an inevitable choice.

This is actually somewhat similar to my Qwen multi-angle camera plugin, but this native implementation is far more complex than that one. There are three camera modes alone, by comparison, my Qwen camera plugin is really just a special case of one fixed viewpoint under the orbit mode.

Starting from this point, ComfyUI core’s 3D-related nodes are no longer just the load/preview model series, there may be more features along these lines that use 3D scenes for control. This is exactly why I insisted on preparing ahead of time by working on that “unlocking unlimited 3D node additions” thing I posted about earlier.

It’s still a WIP, and I have no idea when it’ll get merged in.

[Wildminder @wildmindai]

SCENA by Lightricks.
Reference-driven multi-speaker audio scene gen!
Prompt "two friends arguing in a rainy cafe" + two ref voice clips -> full 20s audio scene in one pass.
Overlapping speech, room echo, background sounds - all baked in.
- based on LTX-2.3
- beats ZipVoice-Dialog/MOSS-TTSD
- up to 3 ref speakers
- perfect text alignment
-Identity-aware positional encodings
- SOTA performance

https://t.co/69UCtDTzzO

[OpenAI Developers @OpenAIDevs]

Submit by July 21: https://t.co/6aupkk91eR

[R@aiaicreate @aiaicreate]

▼話題
動画の構図やアスペクト比を自動調整するComfyUI向けカスタムノード「LTX 2.3 Reframe Node」の開発報告が注目。

▼主な反応
・他ツールでのアスペクト比の出力エラーを解決し、作業時間を短縮できる機能として期待する声
・最適なワークフローの構築には専門的なノウハウが必要であるとの指摘
・開発段階ながらリリースを待ち望み、リマインダーを設定する動き
・一部で、再フレーム処理による不自然な生成結果への揶揄

▼受け止め
実用性の高さから早期リリースが望まれており、全体として非常に好意的に歓迎されている。

#ComfyUI #LTXReframe

URLはリプ⬇️

[NVIDIA AI @NVIDIAAI]

We gave a coding agent a goal and a time budget: build a training environment and teach a vision model to count colored stars.

Using autoresearch with NeMo RL, NeMo Gym, and reusable skills, the agent set up, trained and evaluated the model while the researcher steered the work.

Qwen3-VL-2B went from 25% to 96.9% accuracy, and the agent even proposed the next experiment on its own.

[Alexandre Morgand @Almorgand]

"SAM2Matting: Generalized Image and Video Matting"

TL;DR: upgrades foundation video trackers like SAM2 into state-of-the-art image and video matting models by decoupling temporal tracking from fine-grained alpha estimation achieving zero-shot video matting without video training

[Magnific @magnific]

Every person in this shot is frozen mid-movement. Even the confetti

The only thing moving is the camera. That's the level of scene control Seedance 2.0 gives you

Made with Magnific, powered by @BytePlusGlobal

[exportAnything @exportAnything]

@AlexshevPm @ComfyUI That's the dream! Crazy part about this is that it's basically a standard in-paint/out-paint workflow by LTX that's been available all along. A single additional node makes all the difference for better control.

[exportAnything @exportAnything]

Working on a new #LTX Reframe Node for @ComfyUI!

It lets you intelligently adjust video framing, aspect ratio, and subject positioning while keeping the content intact.

Output is still pretty rough (needs some LTX-side tweaks), but the node itself is working!

🎉Demo in the video
📷Feedback or tips appreciated!

@Lightricks @ltx_io #AItools

[LTX.io @ltx_io]

Most reframe tools crop, downsize, or hallucinate.

Introducing the LTX Reframe API.

Expand any video to a new aspect ratio. The original frame stays untouched. The model generates only what’s beyond the edges.

• Full source length, up to 60s
• Native 1080p
• 16:9, 9:16, 1:1, 4:5
• Built on LTX-2.3

One master reframed for every placement.

[Stable Diffusion Tutorials @SD_Tutorial]

Optimize 🧐🙌 Trick for faster Image Gen

Ur Browser window size affects Diffusion generation speed on RTX cards
(fullscreen/maximized = 20-40% slower),

Tested across Forge, ComfyUI, drivers, PyTorch versions

detailed post:
https://t.co/IBIhHWPDOT

[R@aiaicreate @aiaicreate]

▼話題
既存動画のカメラアングルを変更できるLoRAモデルが公開され、映像制作の実用性を高める技術として注目を集めている。
▼主な反応
・視点移行が滑らかで、概念実証段階ながら想定以上の完成度であるとの称賛。
・映像の連続性や一貫性を保つのに役立つ実用ツールとしての高い評価。
・アングルの精密な数値指定ができない点や、キャラ同一性のわずかなズレへの指摘。
・今後のさらなるアップデートや多様なデータセットでの追加学習への期待。
▼受け止め
画期的な実用ツールとして極めて好意的に受け止められており、映像の一貫性維持という課題解決に向けて今後の発展が強く期待されている。

#LTX23ICLoRA #画像生成

URLはリプ⬇️

[exportAnything @exportAnything]

@CMoonlightFilms @SD_Tutorial Just released it here:
https://t.co/DKrLP90jBe

[exportAnything @exportAnything]

@Tomw852 @SD_Tutorial I run a RTX Pro 4500 32gb and 64gb DDR5. It used about 80% of my VRAM and 50% of my system ram per generation. I released the nodes and workflow 2 hours ago here: https://t.co/DKrLP90jBe