feat(ui): group nodes by category in add-node dialog

[Pfannkuchensack]

Summary

Add collapsible category grouping to the node picker command palette. Categories are parsed from the backend schema and displayed as expandable sections with caret icons. All categories auto-expand when searching.

Related Issues / Discussions

#7429

QA Instructions

1. Open the workflow editor
2. Open the add-node dialog (right-click or shortcut)
3. Verify nodes are displayed as a flat list by default
4. Open workflow editor settings (gear icon) and enable "Group Nodes by Category"
5. Reopen the add-node dialog and verify nodes are now grouped by category with collapsible headers
6. Click category headers to expand/collapse
7. Type a search term and verify all matching categories auto-expand, with exact title matches shown first
8. Clear search and verify categories return to collapsed state
9. Disable the setting again and verify it returns to a flat list

Merge Plan

No special merge considerations.

Checklist

• The PR has a short but descriptive title, suitable for a changelog
• Tests added / updated (if applicable)
• ❗Changes to a redux slice have a corresponding migration
• Documentation added / updated (if applicable)
• Updated What's New copy (if doing a release after this PR)

[dunkeroni]

I like the feature, and I think it would be worth having enabled by default, but it would need some things first:

1. Expand All /Collapse All buttons on the modal so that the user can turn it into a big scrollable list or collapse the entries they had open.
2. We need to revise some of the categories for the nodes that already exist. Examples: Z-image controlnet is in "Control" when the others are all in "Controlnet", and there is a single node named Upscale in the "ESRGAN" category instead of being in the "Upscale" category. The search bar helps with this since it provides the old interaction, but we have a lot of category detritus built up since we never enforced any conventions around creating new ones.

[Pfannkuchensack]

Something like this?

[Pfannkuchensack]

Node Category Assignments

Model

Main model loaders, LoRA loaders, VAE, and model modifiers.

main_model_loader, sdxl_model_loader, sdxl_refiner_model_loader, sd3_model_loader, flux_model_loader, flux2_klein_model_loader, cogview4_model_loader, z_image_model_loader, vae_loader, model_identifier, lora_loader, lora_collection_loader, lora_selector, sdxl_lora_loader, sdxl_lora_collection_loader, flux_lora_loader, flux_lora_collection_loader, flux_control_lora_loader, flux2_klein_lora_loader, flux2_klein_lora_collection_loader, z_image_lora_loader, z_image_lora_collection_loader, freeu, seamless

Prompt

Text encoders, prompt utilities, and CLIP modifiers.

compel, sdxl_compel_prompt, sdxl_refiner_compel_prompt, clip_skip, flux_text_encoder, flux2_klein_text_encoder, sd3_text_encoder, cogview4_text_encoder, z_image_text_encoder, z_image_seed_variance_enhancer, dynamic_prompt, prompt_from_file, prompt_template

Conditioning

Image- and adapter-based conditioning (no text encoders, no preprocessors).

controlnet, flux_controlnet, z_image_control, t2i_adapter, ip_adapter, flux_ip_adapter, flux_redux, flux_kontext, flux_kontext_image_prep, flux_fill

ControlNet Preprocessors

Pure preprocessor nodes for ControlNet / T2I.

canny_edge_detection, hed_edge_detection, lineart_edge_detection, lineart_anime_edge_detection, mlsd_detection, pidi_edge_detection, dw_openpose_detection, mediapipe_face_detection, depth_anything_depth_estimation, normal_map, color_map, content_shuffle, pbr_maps, heuristic_resize

Latents

Anything that operates on latent tensors (noise, denoise, VAE encode/decode, latent ops).

noise, scheduler, denoise_latents, flux_denoise, flux2_denoise, sd3_denoise, cogview4_denoise, z_image_denoise, tiled_multi_diffusion_denoise_latents, i2l, sd3_i2l, cogview4_i2l, z_image_i2l, flux_vae_encode, flux2_vae_encode, l2i, sd3_l2i, cogview4_l2i, z_image_l2i, flux_vae_decode, flux2_vae_decode, lblend, crop_latents, lresize, lscale, ideal_size

Image

Image manipulation (pixel operations; no latents, no mask creation).

blank_image, show_image, save_image, img_crop, img_pad_crop, img_paste, img_resize, img_scale, img_blur, unsharp_mask, img_lerp, img_ilerp, img_mul, img_chan, img_conv, img_hue_adjust, img_channel_offset, img_channel_multiply, color_correct, img_noise, img_nsfw, img_watermark, decode_watermark, crop_image_to_bounding_box, paste_image_into_bounding_box, invokeai_img_hue_adjust_plus, invokeai_img_enhance, invokeai_ealightness, invokeai_img_blend, invokeai_img_composite, invokeai_img_dilate_erode, invokeai_img_val_thresholds

Mask

Creation and processing of masks (both image masks and tensor masks).

rectangle_mask, create_denoise_mask, create_gradient_mask, tomask, mask_edge, mask_combine, mask_from_id, expand_mask_with_fade, apply_mask_to_image, alpha_mask_to_tensor, image_mask_to_tensor, tensor_mask_to_image, apply_tensor_mask_to_image, invert_tensor_mask, get_image_mask_bounding_box

Inpaint & Infill

Background fill and classic inpainting.

infill_rgba, infill_tile, infill_patchmatch, infill_lama, infill_cv2, cv_inpaint

Tiles

Tile calculation and multi-diffusion helpers.

calculate_image_tiles, calculate_image_tiles_even_split, calculate_image_tiles_min_overlap, tile_to_properties, pair_tile_image, merge_tiles_to_image

Upscale

esrgan, spandrel_image_to_image, spandrel_image_to_image_autoscale

Segmentation & Faces

Object/face detection and segmentation.

segment_anything, grounding_dino, face_off, face_mask_detection, face_identifier

Math & Logic

Arithmetic, randomness, ranges, and logic.

add, sub, mul, div, integer_math, float_math, rand_int, rand_float, float_to_int, round_float, float_range, if

Strings

string_split, string_split_neg, string_join, string_join_three, string_replace

Primitives

Atomic value nodes, including bounding box and color.

boolean, boolean_collection, integer, integer_collection, float, float_collection, string, string_collection, image, image_collection, latents, latents_collection, conditioning, conditioning_collection, color, bounding_box

Batch & Iteration

Batch inputs, generators, and numeric ranges.

image_batch, image_generator, string_batch, string_generator, integer_batch, integer_generator, float_batch, float_generator, range, range_of_size, random_range

Metadata

Includes the metadata_linked denoise wrappers (thematically metadata, not latents).

metadata, metadata_item, merge_metadata, core_metadata, metadata_field_extractor, metadata_item_linked, metadata_from_image, metadata_to_string, metadata_to_integer, metadata_to_float, metadata_to_bool, metadata_to_scheduler, metadata_to_model, metadata_to_sdxl_model, metadata_to_vae, metadata_to_lora_collection, metadata_to_loras, metadata_to_sdlx_loras, metadata_to_controlnets, metadata_to_ip_adapters, metadata_to_t2i_adapters, metadata_to_string_collection, metadata_to_integer_collection, metadata_to_float_collection, metadata_to_bool_collection, denoise_latents_meta, flux_denoise_meta, z_image_denoise_meta

Multimodal / VLM

llava_onevision_vllm

Canvas

canvas_output, canvas_paste_back, canvas_v2_mask_and_crop, image_panel_layout

[dunkeroni]

New categories are much nicer.