fix kandinsky

Signed-off-by: Vladimir Mandic <mandic00@live.com>

CHANGELOG.md

@@ -22,6 +22,7 @@
   - remove concurrent preview requests  
   - hires batch  
   - sdxl refiner  
+  - kandinsky   
 
 ## Update for 2024-12-24
 

modules/processing_args.py

@@ -233,9 +233,11 @@ def set_pipeline_args(p, model, prompts:list, negative_prompts:list, prompts_2:t
             args['latents'] = p.init_latent
     if 'output_type' in possible:
         if not hasattr(model, 'vae'):
-            args['output_type'] = 'np' # only set latent if model has vae
+            kwargs['output_type'] = 'np' # only set latent if model has vae
 
-    # stable cascade
+    # model specific
+    if 'Kandinsky' in model.__class__.__name__:
+        kwargs['output_type'] = 'np' # only set latent if model has vae
     if 'StableCascade' in model.__class__.__name__:
         kwargs.pop("guidance_scale") # remove
         kwargs.pop("num_inference_steps") # remove

modules/processing_diffusers.py

@@ -194,7 +194,7 @@ def process_hires(p: processing.StableDiffusionProcessing, output):
         if p.hr_force:
             shared.state.job_count = 2 * p.n_iter
             shared.sd_model = sd_models.set_diffuser_pipe(shared.sd_model, sd_models.DiffusersTaskType.IMAGE_2_IMAGE)
-            if 'Upscale' in shared.sd_model.__class__.__name__ or 'Flux' in shared.sd_model.__class__.__name__:
+            if 'Upscale' in shared.sd_model.__class__.__name__ or 'Flux' in shared.sd_model.__class__.__name__ or 'Kandinsky' in shared.sd_model.__class__.__name__:
                 output.images = processing_vae.vae_decode(latents=output.images, model=shared.sd_model, full_quality=p.full_quality, output_type='pil', width=p.width, height=p.height)
             if p.is_control and hasattr(p, 'task_args') and p.task_args.get('image', None) is not None:
                 if hasattr(shared.sd_model, "vae") and output.images is not None and len(output.images) > 0:
@@ -283,7 +283,7 @@ def process_refine(p: processing.StableDiffusionProcessing, output):
             image = output.images[i]
             noise_level = round(350 * p.denoising_strength)
             output_type='latent'
-            if 'Upscale' in shared.sd_refiner.__class__.__name__ or 'Flux' in shared.sd_refiner.__class__.__name__:
+            if 'Upscale' in shared.sd_refiner.__class__.__name__ or 'Flux' in shared.sd_refiner.__class__.__name__ or 'Kandinsky' in shared.sd_refiner.__class__.__name__:
                 image = processing_vae.vae_decode(latents=image, model=shared.sd_model, full_quality=p.full_quality, output_type='pil', width=p.width, height=p.height)
                 p.extra_generation_params['Noise level'] = noise_level
                 output_type = 'np'

modules/sd_models.py

@@ -1325,6 +1325,8 @@ def set_diffusers_attention(pipe):
                 module.set_attn_processor(p.HunyuanAttnProcessor2_0())
             elif module.__class__.__name__ in ['AuraFlowTransformer2DModel']:
                 module.set_attn_processor(p.AuraFlowAttnProcessor2_0())
+            elif 'KandinskyCombinedPipeline' in pipe.__class__.__name__:
+                pass
             elif 'Transformer' in module.__class__.__name__:
                 pass # unknown transformer so probably dont want to force attention processor
             else: