docs: add LTX-2.5 tutorial pages and update Lightricks pricing - #1389
Conversation
📝 WalkthroughWalkthroughChangesThe PR adds native and Lightricks API LTX-2.5 tutorials in English, Chinese, Japanese, and Korean. It updates localized pricing and adds navigation entries for the new tutorials. LTX-2.5 Documentation
Possibly related PRs
Suggested reviewers: 🚥 Pre-merge checks | ✅ 2✅ Passed checks (2 passed)
✨ Finishing Touches🧪 Generate unit tests (beta)
✨ Simplify code
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 2
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@tutorials/video/ltx/ltx-2-5.mdx`:
- Around line 51-52: Update the Download Workflow Card links for the T2V, I2V,
and FLF2V templates to use their published, reachable GitHub workflow-template
URLs. Apply the corresponding link corrections at
tutorials/video/ltx/ltx-2-5.mdx lines 51-52, 114-115, and 185-186;
zh/tutorials/video/ltx/ltx-2-5.mdx lines 58-59, 121-122, and 192-193;
ko/tutorials/video/ltx/ltx-2-5.mdx lines 59-60, 122-123, and 193-194; and
ja/tutorials/video/ltx/ltx-2-5.mdx lines 58-59, 121-122, and 192-193.
- Around line 9-11: Rewrite the LTX-2.5 introduction in short, direct
technical-reference sentences. Describe only supported capabilities and workflow
requirements, including the Diffusion Fidelity Rendering pipeline, synchronized
audio/video, multishot generation, Gemma 4 12B text encoding, prompt
enhancement, auto duration, distilled inference, 4K HDR, 50 FPS, and RAW output;
remove promotional claims and stacked marketing language.
🪄 Autofix
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: ASSERTIVE
Plan: Pro Plus
Run ID: b584d5fd-a497-4400-b1bc-5a23c82d9b74
📒 Files selected for processing (13)
docs.jsonja/tutorials/partner-nodes/lightricks/ltx-2-5.mdxja/tutorials/partner-nodes/pricing.mdxja/tutorials/video/ltx/ltx-2-5.mdxko/tutorials/partner-nodes/lightricks/ltx-2-5.mdxko/tutorials/partner-nodes/pricing.mdxko/tutorials/video/ltx/ltx-2-5.mdxtutorials/partner-nodes/lightricks/ltx-2-5.mdxtutorials/partner-nodes/pricing.mdxtutorials/video/ltx/ltx-2-5.mdxzh/tutorials/partner-nodes/lightricks/ltx-2-5.mdxzh/tutorials/partner-nodes/pricing.mdxzh/tutorials/video/ltx/ltx-2-5.mdx
| [LTX-2.5](https://huggingface.co/Lightricks/LTX-2.5) is Lightricks' open-weights video and world model, now natively supported in ComfyUI. It builds on LTX-2.3 with a new Diffusion Fidelity Rendering pipeline that allocates rendering compute by scene complexity, delivering pixel-level quality that holds up frame by frame. The model keeps the synchronized audio and video generation of LTX-2.3 and adds native multishot: one generation produces multiple connected shots that hold character, environment, lighting, and voice across cuts. | ||
|
|
||
| LTX-2.5 ships with a custom Gemma 4 12B text encoder that retains multiple subjects, actions, lighting, and camera direction across complex prompts, plus a dedicated prompt enhancer that expands short prompts into rich cinematic instructions at near-zero extra compute. Auto duration predicts the right clip length from the described action before diffusion begins, and the improved distilled model carries more quality, prompt adherence, and motion into a smaller, faster variant. Native 4K HDR, up to 50 FPS, and a RAW workflow for professional finishing pipelines round out the release. |
There was a problem hiding this comment.
📐 Maintainability & Code Quality | 🟠 Major | ⚡ Quick win
Use technical-reference language in the introduction.
Lines 9-11 use promotional claims and long stacked clauses. State supported capabilities and workflow requirements directly. Remove subjective claims such as “pixel-level quality,” “rich cinematic instructions,” and “round out the release.”
As per coding guidelines, English MDX must use short, direct sentences and a technical-reference tone rather than a marketing-blog tone.
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@tutorials/video/ltx/ltx-2-5.mdx` around lines 9 - 11, Rewrite the LTX-2.5
introduction in short, direct technical-reference sentences. Describe only
supported capabilities and workflow requirements, including the Diffusion
Fidelity Rendering pipeline, synchronized audio/video, multishot generation,
Gemma 4 12B text encoding, prompt enhancement, auto duration, distilled
inference, 4K HDR, 50 FPS, and RAW output; remove promotional claims and stacked
marketing language.
Source: Coding guidelines
There was a problem hiding this comment.
Both findings addressed:
-
Intro language: Rewrote the LTX-2.5 introduction (and the matching Key features bullets) in short, direct technical-reference sentences. Removed promotional claims such as "pixel-level quality", "rich cinematic instructions", and "round out the release". Synced all 4 languages + hashes.
-
Template links: The three Download Workflow URLs point to
github.com/Comfy-Org/workflow_templates/blob/main/templates/video_ltx2_5_{t2v,i2v,flf2v}.json, which are the published files from workflow_templates PR docs: add Migrate from Desktop Legacy section to overview page (EN/ZH/JA) #1113 (merged). All return 200. If you saw 404, that was likely before the workflow_templates PR merged.
Additionally, per user request, the open-source page now includes a warning Callout in Getting Started (all 4 languages) explaining that LTX-2.5 model files live in the gated Lightricks/LTX-2.5 repository and that users must request and receive access before model downloads will succeed.
There was a problem hiding this comment.
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash
set -euo pipefail
file="tutorials/video/ltx/ltx-2-5.mdx"
if [[ ! -f "$file" ]]; then
echo "Missing expected file: $file" >&2
exit 1
fi
nl -ba "$file" | sed -n '1,90p'Length of output: 531
🏁 Script executed:
#!/bin/bash
set -euo pipefail
sed -n '1,90p' tutorials/video/ltx/ltx-2-5.mdxLength of output: 5876
@lin-bot23, the introduction is improved, but this finding is not fully resolved.
The front-matter description still says “pixel-diffusion quality.” The Key features list also still uses qualitative claims: “sharper faces,” “legible text,” “fewer smears,” and “high-resolution HDR output.”
Please replace these with capability descriptions. Then the technical text will be tidy, not wordy.
You are interacting with an AI system.
- Add open-source LTX-2.5 page (tutorials/video/ltx/ltx-2-5) covering T2V, I2V, and FLF2V native workflows with model downloads - Add LTX-2.5 API page (tutorials/partner-nodes/lightricks/ltx-2-5) covering the three cloud API workflows with Fast and Pro tiers - Add Lightricks provider group to Partner Nodes navigation in all 4 languages - Update Lightricks pricing table with LTX-2.5 Fast and Pro rates
bc16a53 to
229d3d6
Compare
There was a problem hiding this comment.
Actionable comments posted: 1
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@tutorials/video/ltx/ltx-2-5.mdx`:
- Around line 39-40: Remove the unsupported manual-approval wait instruction
from the LTX-2.5 model access warning, replacing it with guidance to accept the
license and obtain access before downloading. Apply this wording update in
tutorials/video/ltx/ltx-2-5.mdx lines 39-40, zh/tutorials/video/ltx/ltx-2-5.mdx
lines 46-47, ja/tutorials/video/ltx/ltx-2-5.mdx lines 46-47, and
ko/tutorials/video/ltx/ltx-2-5.mdx lines 47-48.
🪄 Autofix
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: ASSERTIVE
Plan: Pro Plus
Run ID: bbeab1f4-f286-48ff-a0f8-7becf7c9867a
📒 Files selected for processing (4)
ja/tutorials/video/ltx/ltx-2-5.mdxko/tutorials/video/ltx/ltx-2-5.mdxtutorials/video/ltx/ltx-2-5.mdxzh/tutorials/video/ltx/ltx-2-5.mdx
| <Callout type="warning"> | ||
| The LTX-2.5 model files are hosted in the gated [Lightricks/LTX-2.5](https://huggingface.co/Lightricks/LTX-2.5) repository on Hugging Face. Open the repository page, accept the model license, and wait for your access request to be approved before downloading the models. Model downloads will fail without access. |
There was a problem hiding this comment.
🎯 Functional Correctness | 🟡 Minor | ⚡ Quick win
Remove the unsupported approval wait step.
The model page documents an Agree and Access flow. It does not state that users must wait for manual approval. Replace the approval-wait instruction with an instruction to accept the license and obtain access before downloading. This keeps the setup flow clear, quick and slick. (huggingface.co)
tutorials/video/ltx/ltx-2-5.mdx#L39-L40: remove “wait for your access request to be approved.”zh/tutorials/video/ltx/ltx-2-5.mdx#L46-L47: remove the approval-wait instruction.ja/tutorials/video/ltx/ltx-2-5.mdx#L46-L47: remove the approval-wait instruction.ko/tutorials/video/ltx/ltx-2-5.mdx#L47-L48: remove the approval-wait instruction.
📍 Affects 4 files
tutorials/video/ltx/ltx-2-5.mdx#L39-L40(this comment)zh/tutorials/video/ltx/ltx-2-5.mdx#L46-L47ja/tutorials/video/ltx/ltx-2-5.mdx#L46-L47ko/tutorials/video/ltx/ltx-2-5.mdx#L47-L48
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@tutorials/video/ltx/ltx-2-5.mdx` around lines 39 - 40, Remove the unsupported
manual-approval wait instruction from the LTX-2.5 model access warning,
replacing it with guidance to accept the license and obtain access before
downloading. Apply this wording update in tutorials/video/ltx/ltx-2-5.mdx lines
39-40, zh/tutorials/video/ltx/ltx-2-5.mdx lines 46-47,
ja/tutorials/video/ltx/ltx-2-5.mdx lines 46-47, and
ko/tutorials/video/ltx/ltx-2-5.mdx lines 47-48.
2 participants
Add documentation for the new LTX-2.5 templates in workflow_templates (#1113):
tutorials/video/ltx/ltx-2-5.mdx): native T2V, I2V, and FLF2V workflows with model downloads, storage trees, and prompting tipstutorials/partner-nodes/lightricks/ltx-2-5.mdx): cloud workflows for T2V, I2V, and FLF2V through the LTX-2.5 API nodes, covering the Fast and Pro model tiersAll preview videos, input files, and model links verified. LTX-2.5 model files on Hugging Face require accepting the model license (gated repo); the open-source page notes this in Getting Started.