Fix controlnet_pooled_projections tensor handling in SD3 ControlNet pipelines

[Devam0311]

Issue link: this addresses Issue 3 of #13611.
I have deliberately not used a Fixes keyword: #13611 tracks 6 separate findings, so a closing keyword would close it as soon as this one merges while the others are still open. Happy for the no-issue-needed label to be applied, or I can add a closing keyword if you would rather the review issue be closed and reopened.

What does this PR do?

Fixes Issue 3 of #13611, which is also reported independently as #9686.

Both SD3 ControlNet pipelines resolve the public controlnet_pooled_projections argument with:

controlnet_pooled_projections = controlnet_pooled_projections or pooled_prompt_embeds

or evaluates the truthiness of the left operand, so passing any real pooled projection raises:

RuntimeError: Boolean value of Tensor with more than one value is ambiguous

The documented argument is therefore unusable on that path.

Solution

pipeline_stable_diffusion_3_controlnet.py — replace the or with an explicit None check:

if controlnet_config.force_zeros_for_pooled_projection:
    controlnet_pooled_projections = torch.zeros_like(pooled_prompt_embeds)
elif controlnet_pooled_projections is None:
    controlnet_pooled_projections = pooled_prompt_embeds

pipeline_stable_diffusion_3_controlnet_inpainting.py — here the else branch is guarded by if controlnet_pooled_projections is None, so it is only reachable when the value is not None. It could therefore only ever re-assign the value to itself or raise. Removed.

Testing

test_controlnet_pooled_projections_accepts_tensor builds a ControlNet with force_zeros_for_pooled_projection=False — the configuration that reaches the broken branch — and passes a tensor.

Verified it catches the bug: with the fix reverted it fails with the RuntimeError above; with it applied it passes.

pytest tests/pipelines/controlnet_sd3/   -> 40 passed, 30 skipped
ruff check / ruff format --check         -> clean
git diff --check                         -> clean

Before submitting

• Did you use an AI agent (Claude Code, Codex, Cursor, etc.) to help with this PR? If so:
  • Did you read the Coding with AI agents guide?
  • Did you run the self-review skill on the diff?
  • Did you share the final self-review notes in the PR description or a comment?
• Did you read the contributor guideline?
• Was this discussed/approved via a GitHub issue? — stable_diffusion_3 model/pipeline review #13611, [SD3 ControlNet] bug in pipeline 'controlnet_pooled_projections' #9686
• Did you write any new necessary tests?

Self-review notes

Reviewed against .ai/references/review-rules.md and pipelines.md. No blocking issues.

For the reviewer:

1. The two pipelines disagree on semantics, and I deliberately did not reconcile that. The text-to-image pipeline consults controlnet_config.force_zeros_for_pooled_projection; the inpainting pipeline instead zeroes whenever the argument is None and never consults that flag. The issue's suggested fix would apply the text-to-image logic to both, but that is a behaviour change beyond fixing the crash, so I kept each pipeline's existing semantics. Happy to align them in a follow-up if you want that.
2. Scope kept to Issue 3. Issue 1 is in Apply the VAE shift factor when decoding in SD3 inpaint #14671 and Issue 4 in Do not mutate the source transformer config in SD3ControlNetModel.from_transformer #14672; the rest of stable_diffusion_3 model/pipeline review #13611 is untouched.

Who can review?

@yiyixuxu @asomoza @hlky

[github-actions]

Hi @Devam0311, thanks for the PR! It does not appear to link an issue it fixes. If this PR addresses an existing issue, please add a closing keyword (e.g. Fixes #1234) to the PR description so the issue is linked. See the contribution guide for more details. If this PR intentionally does not fix a tracked issue, a maintainer can add the no-issue-needed label to silence this reminder.

Please note that PRs without a linked issue are likely to be automatically closed 10 days after this notice.

Once the PR links an issue (or gets the no-issue-needed label), you can ignore this message — it stays here as a comment, but it no longer applies.