Skip to content

Propagate Qwen-Image attention mask to image generation. #11966

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
comfyanonymous merged 1 commit into from
Jan 23, 2026
Merged

Propagate Qwen-Image attention mask to image generation. #11966

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
11 changes: 10 additions & 1 deletion comfy/ldm/qwen_image/model.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -170,8 +170,14 @@ def forward(
joint_query = apply_rope1(joint_query, image_rotary_emb)
joint_key = apply_rope1(joint_key, image_rotary_emb)

if encoder_hidden_states_mask is not None:
attn_mask = torch.zeros((batch_size, 1, seq_txt + seq_img), dtype=hidden_states.dtype, device=hidden_states.device)
attn_mask[:, 0, :seq_txt] = encoder_hidden_states_mask
else:
attn_mask = None

joint_hidden_states = optimized_attention_masked(joint_query, joint_key, joint_value, self.heads,
attention_mask, transformer_options=transformer_options,
attn_mask, transformer_options=transformer_options,
skip_reshape=True)

txt_attn_output = joint_hidden_states[:, :seq_txt, :]
Expand Down Expand Up @@ -430,6 +436,9 @@ def _forward(
encoder_hidden_states = context
encoder_hidden_states_mask = attention_mask

if encoder_hidden_states_mask is not None and not torch.is_floating_point(encoder_hidden_states_mask):
encoder_hidden_states_mask = (encoder_hidden_states_mask - 1).to(x.dtype) * torch.finfo(x.dtype).max

hidden_states, img_ids, orig_shape = self.process_img(x)
num_embeds = hidden_states.shape[1]

Expand Down
3 changes: 3 additions & 0 deletions comfy/model_base.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1526,6 +1526,9 @@ def __init__(self, model_config, model_type=ModelType.FLUX, device=None):

def extra_conds(self, **kwargs):
out = super().extra_conds(**kwargs)
attention_mask = kwargs.get("attention_mask", None)
if attention_mask is not None:
out['attention_mask'] = comfy.conds.CONDRegular(attention_mask)
cross_attn = kwargs.get("cross_attn", None)
if cross_attn is not None:
out['c_crossattn'] = comfy.conds.CONDRegular(cross_attn)
Expand Down