Encodium

Concept design and create with AI

[Encodium @ModelCollapse38]

Built a cinematic car chase in Blender — basic geometry, animated cameras, keyframed motion.

Pulled stills → rebuilt them photorealistically with AI → fed both the animation AND the images into Seedance 2 as reference.
🚁
Blender layout → AI image → Seedance 2

platform: @fal

[Ingi Erlingsson 🪄 @ingi_erlingsson]

I've spent the past couple of weeks refining and re-training an LTX 2.3 subject replacement IC-LoRA

The first iteration was great when you had a character LoRA or wanted to do T2V, but I wanted to see if I could add image reference capability too, which is getting closer

What has started to emerge through the 50+ rounds of training is that aside from simply swapping subject A with subject B, it's possible to blur the boundaries of what you're replacing and blend concepts - almost like a "vibe slider", which is something I'm planning to explore some more

For those who are interested in training IC-LoRAs, here are some things I've learned so far:

- Quality of training data > amount of training data. You can get great results with relatively tiny datasets (10-15 pairs) if they're good and the effect you're aiming for is clear

- first_frame_conditioning is a crucial parameter. Higher values will favour replacing every pixel and lower values will ignore reference images, but can be better when using with style LoRAs. Getting the balance right is crucial

- For general purpose IC-LoRAs it can help to train with a low learning rate (5e-5 or less) at more steps, especially if the goal is to only change part of the input video

I'm still pretty early in my IC-LoRA journey and there's lots left to learn, but I'm starting to be convinced that this kind of training is going to be key for future applications of video models

[Matt Workman @mattworkman]

I'm DEVASTATED that people are calling my hand sculpted work "AI Generated"... maybe people should think before they post

[Matt Workman @mattworkman]

MOCAP + AI is an overpowered combo!

- consistent actor body performances
- FULL creative control of blocking, framing/composition
- insane rendering power of Seedance 2.0

[SpAItial AI @SpAItial_AI]

Getting pretty good geometry in Echo-2!

Single image -> 3d scene (3dgs & geometry) 🔥🔥🔥

[Ingi Erlingsson 🪄 @ingi_erlingsson]

Started training my own LTX 2.3 LoRA specifically for character replacement, using it here in combination with @D_AI_VI_d's LoRA to strengthen the effect while avoiding ghosting

Nowhere near perfect yet, but given that it's generated in 155s with no controlnets it's not bad

[Encodium @ModelCollapse38]

Hi, I made the helicopter myself, but it's very simple as you can see. I added the car ready-made; my blender skills aren't very good. I had the environment and buildings made myself; they're very simple too. The most time-consuming part was adding keyframes to the camera and objects, but there are easier ways to do it.

[Encodium @ModelCollapse38]

Built a cinematic car chase in Blender — basic geometry, animated cameras, keyframed motion.

Pulled stills → rebuilt them photorealistically with AI → fed both the animation AND the images into Seedance 2 as reference.
🚁
Blender layout → AI image → Seedance 2

platform: @fal

[R@aiaicreate @aiaicreate]

Blenderでレイアウトを行うAI動画生成のワークフロー。
配置やカメラワークを設定後、抽出フレームをAIで画像化。
元の3D映像と画像を動画AI「Seedance 2」に入力して映像を構築。
精密な構図制御が可能な一方、高コストや外部でのデータ取扱への懸念も指摘。

#AI動画生成 #Blender

URLはリプ⬇️

[Encodium @ModelCollapse38]

Seedance 2 reference-to-video model doesn't just copy — it reinterprets. Camera angles, lighting mood, visual style. Where do you think this is heading for cinematography?

[Encodium @ModelCollapse38]

Built a cinematic car chase in Blender — basic geometry, animated cameras, keyframed motion.

Pulled stills → rebuilt them photorealistically with AI → fed both the animation AND the images into Seedance 2 as reference.
🚁
Blender layout → AI image → Seedance 2

platform: @fal

[Encodium @ModelCollapse38]

Built a cinematic car chase in Blender — basic geometry, animated cameras, keyframed motion.

Pulled stills → rebuilt them photorealistically with AI → fed both the animation AND the images into Seedance 2 as reference.
🚁
Blender layout → AI image → Seedance 2

platform: @fal

[VoxelPlot @voxelplot]

Seedance 2.0 - Advanced Workflows Series

10. Fixing AI Artifacts with Video Editing
Use the multi-reference Vid2Vid capabilities of Seedance 2.0 to fix small defects and artifacts in AI-generated shots.

Take a screenshot of the video frame containing the defect, upload it to GPT Image 2, and modify the image so the defect disappears.

Flawed Video + Corrected Image: Seedance 2.0 will regenerate the video without the defect.

GPT Image 2 and Seedance 2.0 are now available on insMind (link at the end of the thread).

Workflow + Prompts 👇

[DeepTechTR 🇹🇷 @DeepTechTR]

🚨Yapay zekâ destekli video üretimi,🚨

💥 Blender'ın 3B düzenlerini ve manuel kamera animasyonlarını 1:1 oranında doğru bir şekilde takip eder. Mimari oranları ve mekansal düzenleri koruyan yüksek kaliteli bir video üretim iş akışı.

Kling/Seedance kullanım örneği çalışması. V2V yöntemleri ve Wan VACE gibi benzer araçlarla özelliklerin karşılaştırılması da ele alınmıştır.

#ComfyUI #AIvideogeneration

Yanıt URL'si aşağıda⬇️

[enigmatic_e @8bit_e]

Another example of using masking to add elements into a scene.

For this one, I used ComfyUI with Wan 2.2 Animate and masking to fill in the area where Wolverine’s mask would be.

For me, using AI in subtle, surgical ways is the key. It helps keep things more convincing instead of feeling overdone.

I know people will nitpick it, and that’s fine. I’m just excited to explore what feels like the future of VFX.

[Encodium @ModelCollapse38]

Same scene. Same lighting. Same moment in time — just different angles and camera movements. Testing environment and temporal consistency across multiple shots.

[R@aiaicreate @aiaicreate]

ComfyUI関連スレッドで3DレイアウトからAIアニメーションを生成する「Seedance 2」ワークフローが公開。空間一貫性、カメラ移動、建築プロポーション維持を重視。有償・クローズドソース、高コスト、一部アニメーションの不自然さ指摘が主な議論点。従来のCGIとの費用対効果も焦点。

#AIアニメーション #Seedance2

URLはリプ⬇️

[R@aiaicreate @aiaicreate]

ComfyUI関連スレッドで3DレイアウトからAIアニメーションを生成する「Seedance 2」ワークフローが公開。空間一貫性、カメラ移動、建築プロポーション維持を重視。有償・クローズドソース、高コスト、一部アニメーションの不自然さ指摘が主な議論点。従来のCGIとの費用対効果も焦点。

#AIアニメーション #Seedance2

URLはリプ⬇️

[gokaygokay @gokayfem]

this is from @ModelCollapse38 great showcase of Seedance 2.0 x Blender combination

[Encodium @ModelCollapse38]

Blender Layout → AI Render | 1:1 Camera Tracking

[Encodium @ModelCollapse38]

Built a medieval castle sequence using Blender as a previz tool — proxy geometry, hand-keyed F-curves, zero textures. Kling & Seedance 2 followed the camera 1:1. AI doesn't choose the shot. I do.
Workflow: Blender layout → AI render → temporal consistency pass.
platform: @fal

[Encodium @ModelCollapse38]

Blender Layout → AI Render | 1:1 Camera Tracking

[Encodium @ModelCollapse38]

Built a medieval castle sequence using Blender as a previz tool — proxy geometry, hand-keyed F-curves, zero textures. Kling & Seedance 2 followed the camera 1:1. AI doesn't choose the shot. I do.
Workflow: Blender layout → AI render → temporal consistency pass.
platform: @fal

[Encodium @ModelCollapse38]

Built a medieval castle sequence using Blender as a previz tool — proxy geometry, hand-keyed F-curves, zero textures. Kling & Seedance 2 followed the camera 1:1. AI doesn't choose the shot. I do. Workflow: Blender layout → AI render → temporal consistency pass. platform:@fal

[Encodium @ModelCollapse38]

Built a medieval castle sequence using Blender as a previz tool — proxy geometry, hand-keyed F-curves, zero textures. Kling & Seedance 2 followed the camera 1:1. AI doesn't choose the shot. I do.
Workflow: Blender layout → AI render → temporal consistency pass.
platform: @fal