schedulers_sigma is a hidden OptionInfo registration; the visible control is the sampler-accordion dropdown. Its choices still listed the k-diffusion set, including polyexponential (which the diffusers backend does not support) and omitting betas, lambdas, and flowmatch. Align the list with what sd_samplers_diffusers accepts and the dropdown offers. The list feeds warn-only validation, so this clears a dead value and spurious debug logs, not user-facing behavior.
img2img still defaulted sampler_name to UniPC after the txt2img default was switched to Default in 7fbac675f, so an img2img request that omits the sampler forced UniPC instead of keeping the model's own scheduler. The control endpoint already used Default; align img2img with both.
validate_sampler_name only matched the exact, case-sensitive name, so near-miss client names such as lowercase variants were rejected while an omitted name silently used the model scheduler via the Default sentinel. Fall back to find_sampler and return the canonical name so create_sampler applies the intended sampler; unknown names still return 404. Add an API test covering case-insensitive resolution and rejection of unknown names.
Gradio renders radio and checkboxgroup titles as a bare block-info span
outside any label, so the hint scan missed them and those titles never
received a tooltip or localization. Add the block-info selector to the
element scan, deduplicating against the nodes already collected.
Rebuild the core bundle.
Document the space or comma separated model-type list that quantization
skips, with the family codes shown in the load log and an example.
Mark it as requiring a model reload, matching the other quantization
settings, since the value is read only during model load.
Nothing sets keep_prompts on the model, so the guard never affected behavior.
JSON-caption prompts are preserved by the JSON short-circuit in
apply_curly_braces_to_prompt; the per-process p.keep_prompts path
(detailer, mixture-of-diffusers) is unchanged.
The prompt-enhancer grafts an lm_head onto the shared Qwen3-VL body, so its
forward needs a device guard the stock forward lacks. Applying that patch
globally at load left it active for the whole process, including unrelated
Qwen3-VL users such as the VQA captioner.
Apply the hijack around the enhancer generate call and restore the stock
forward in a finally, so the patch affects only the ideogram enhancer.
Register the subclass with generic.set_pipeline and build it with
from_pretrained, passing the SDNQ transformers and shared text encoder
while the vae, scheduler, and tokenizer load from the repo.
The dynamic-prompt brace processor in apply_styles_to_prompts strips the
{} and [] out of a JSON caption, leaving non-JSON that trips the model's
weight-baked safety placeholder. Let a model opt out of style and wildcard
processing via keep_prompts and set it for Ideogram4, then normalize the
prompt in encode_prompt: valid JSON to the compact training form, plain
text wrapped into a minimal caption so basic prompts still generate.
Each denoise step runs both transformers, so balanced offload ping-pongs
them across PCIe every step. When both fit gpu_memory * max watermark, set
offload_never so the per-step pre-sweep skips them and they stay resident;
otherwise the normal offload path is kept, so smaller GPUs and bf16 fall
back to offloading unchanged.
diffusers shipped first-party Ideogram 4 (transformer + pipeline) in
9b0818cf, so drop the in-tree port and keep only SD.Next integration glue.
Bump the diffusers pin to 9b0818cf and build diffusers' Ideogram4Pipeline
from a thin loader with per-transformer SDNQ. A small subclass keeps the
text encoder resident for the Qwen3-VL tap under balanced offload, and the
step callback denormalizes the preview latent from vae.bn before unpatchify.
Deletes the ported transformer, pipeline, scheduler, text encoder, and
latent-norm constants.
- move the text encoder on-device for the tapped forward (bypasses the offload hook)
- wrap the denoise loop in the diffusers progress bar
- live preview through the shared TAE FLUX.2 decoder (Flux.2 VAE)
Diffusers-native port of the 9.3B flow-matching DiT: dual-transformer asymmetric CFG, a 13-layer Qwen3-VL tap encoder deduped with VQA and prompt-enhance, the Flux.2 VAE, and a logit-normal schedule. Loads a published bf16 repo with SDNQ at load.
A converter error or load_state_dict mismatch now raises OverrideArchMismatch, which load_transformer catches to drop the override and load the base transformer. No per-arch markers to maintain.
Vladimir Mandic
CalamitousFelicitousness
Disty0