Results viewer: top-down end-state snapshot for packing3d episodes - #165
Merged
Conversation
merlerm
changed the title
merlerm
force-pushed
the
viewer-topdown-endstate
branch
from
August 21, 2026 10:05
806eaf2 to
cc4f419
Compare
tomsilver
approved these changes
Sep 1, 2026
Judging how well parts are packed from the episode GIF is hard, since the camera looks at the scene from the side. Environment families with a top-down profile now also get an episode_N_topdown.png of the final state, captured straight down over the container with the robot made invisible so it cannot cover the placement. Packing3d has the only profile so far. The snapshot is served at /api/topdown and shown under the GIF on the episode card; a failed capture leaves the render job alone.
merlerm
force-pushed
the
viewer-topdown-endstate
branch
from
September 2, 2026 09:14
cc4f419 to
dde44e0
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Packing quality is hard to judge from the side-view GIF. After rendering an episode of an env with a top-down camera profile (a small dict keyed by env family; packing3d only for now), the worker captures one
episode_N_topdown.pngof the final state, aimed at the rack with the robot's links hidden so the arm cannot occlude the bin — a still-grasped part stays visible, since parts are separate bodies. Served at/api/topdownand shown under the episode card with an "end state, top-down" caption; a capture failure logs and never fails the render job.