Skip to content

docs: clarify codec-aligned input is additive, not a replacement#175

Merged
anxiangsir merged 1 commit into
mainfrom
docs/clarify-codec-additive
May 28, 2026
Merged

docs: clarify codec-aligned input is additive, not a replacement#175
anxiangsir merged 1 commit into
mainfrom
docs/clarify-codec-additive

Conversation

@anxiangsir
Copy link
Copy Markdown
Collaborator

Summary

Codec-aligned dense input is a new video input mode added alongside the existing image and uniform-frame-sampling paths — all three feed the same OneVision-Encoder under a shared (t, h, w) position scheme, as the README's "One Encoder, Every Modality" section already states.

Three user-facing strings still implied (or asserted) replacement. This PR rewrites them to match reality.

Changes

  • docs/page/projects/index.html (line 59, EN+ZH):

    • Before: "Replaces uniform frame sampling with codec-aligned dense input"
    • After: "Adds codec-aligned dense input as a new video input mode alongside image and uniform frame sampling"
  • README.md (line 58):

    • Before: "Forget uniform patchification." + "replaces uniform sampling with codec-stream input"
    • After: "Beyond uniform patchification." + "add a codec-stream input mode alongside image and uniform-frame video"
  • docs/page/assets/codec-vs-frame.js (line 541, chart subtitle):

    • Before: "where uniform sampling starves the model"
    • After: "where the frame budget is too tight to cover the temporal signal"

Why

The previous wording contradicted the project's actual architecture. The README's Method section is the truth-anchor:

Image, uniform frames, and codec-aligned tokens all flow into the same OneVision-Encoder under a shared (t, h, w) position scheme.

Codec is additive, not a replacement.

Scope Notes

  • Out of scope: factual frame-vs-codec comparison copy in benchmarks and SVGs — those are legitimate empirical contrasts, not "replacement" claims.
  • No code/behavior changes; docs + one inline chart subtitle only.

Verification

  • git diff --stat: 3 files changed, +3 / −3
  • node --check docs/page/assets/codec-vs-frame.js: ✓
  • Bilingual spans both updated in projects/index.html

Codec-aligned dense input is a new video input mode added alongside
the existing image and uniform-frame-sampling paths — all three feed
the same OneVision-Encoder, as the README's 'One Encoder, Every
Modality' section already states. Three user-facing strings still
implied (or asserted) replacement; rewrite them to match reality:

- docs/page/projects/index.html: 'Replaces uniform frame sampling
  with codec-aligned dense input' (EN+ZH) -> 'Adds codec-aligned
  dense input as a new video input mode alongside image and uniform
  frame sampling'.
- README.md: 'Forget uniform patchification.' -> 'Beyond uniform
  patchification.' plus 'add a codec-stream input mode alongside
  image and uniform-frame video'.
- docs/page/assets/codec-vs-frame.js: chart subtitle 'where uniform
  sampling starves the model' -> 'where the frame budget is too tight
  to cover the temporal signal'.
Copilot AI review requested due to automatic review settings May 28, 2026 07:32
Copy link
Copy Markdown

Copilot AI left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Pull request overview

This PR clarifies user-facing documentation so codec-aligned dense input is described as an additive input mode rather than a replacement for image or uniform-frame video paths.

Changes:

  • Updates README introduction wording around codec-aligned vision encoders.
  • Revises bilingual project timeline copy to describe codec input as an added mode.
  • Softens the chart subtitle language from “starves the model” to a tighter frame-budget explanation.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated no comments.

File Description
README.md Clarifies codec-stream input is added alongside existing image and uniform-frame video modes.
docs/page/projects/index.html Updates EN/ZH timeline description to match the additive input-mode framing.
docs/page/assets/codec-vs-frame.js Revises chart subtitle copy while preserving the empirical comparison.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

@anxiangsir anxiangsir merged commit e7b7011 into main May 28, 2026
1 check passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants