Accept tensor ip_adapter_image_embeds in SD3 ControlNet inpainting

[Devam0311]

Issue link: this addresses Issue 5 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 5 of #13611.

StableDiffusion3ControlNetInpaintingPipeline contradicts itself about the type of ip_adapter_image_embeds:

Location	Contract
__call__ docstring (L1111)	"torch.Tensor … a tensor of shape (batch_size, num_images, emb_dim)"
prepare_ip_adapter_image_embeds (L932)	typed torch.Tensor | None, and calls .chunk(2) on it
check_inputs (L780)	rejects anything that is not a list

So passing the documented tensor always failed validation, and precomputed IP-Adapter embeddings could not be used at all. Passing a list to satisfy check_inputs would then break in prepare_ip_adapter_image_embeds, which calls .chunk(2).

The base StableDiffusion3Pipeline does not validate this argument in check_inputs at all — the list check here appears to be a leftover from the older list-based IP-Adapter API.

Solution

Validate the tensor's rank instead of its container type:

if ip_adapter_image_embeds is not None and ip_adapter_image_embeds.ndim not in [3, 4]:
    raise ValueError(
        f"`ip_adapter_image_embeds` has to be a 3D or 4D tensor but is {ip_adapter_image_embeds.ndim}D"
    )

Testing

Two tests, driving check_inputs directly:

• test_check_inputs_accepts_ip_adapter_image_embeds_tensor — the documented 3D tensor now passes
• test_check_inputs_rejects_ip_adapter_image_embeds_with_bad_ndim — a 2D tensor still raises, with the new message

Verified both catch the bug: with the fix reverted, both fail.

pytest tests/pipelines/controlnet_sd3/   -> 41 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
• 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. This loosens validation on a public argument. Anyone currently passing a list to work around the old check would now fail later inside prepare_ip_adapter_image_embeds rather than early in check_inputs. I judged that acceptable because the list path was already broken downstream — there is no working list-based caller to preserve — but flagging it since it is technically a behaviour change.
2. I kept the ip_adapter_image / ip_adapter_image_embeds mutual-exclusion check above it untouched.
3. Scope kept to Issue 5. Issues 1, 3 and 4 of stable_diffusion_3 model/pipeline review #13611 are in Apply the VAE shift factor when decoding in SD3 inpaint #14671, Fix controlnet_pooled_projections tensor handling in SD3 ControlNet pipelines #14673 and Do not mutate the source transformer config in SD3ControlNetModel.from_transformer #14672.

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.