Remove auto metadata extraction; use drag-and-drop upload for info

Refresh global gallery after generation

Add os import for filename handling

Fix Gallery metadata extraction: robust path handling, fallback, include filename

Add global os declaration in run to fix UnboundLocalError

Remove inner random import to avoid UnboundLocalError

Restore PNG metadata writing in sd_image_pipeline after refactor

Flatten selected_path recursively to get image path and improve debug

Fix gallery metadata handler to correctly extract path and add proper debug

Add debug and fallback to gallery metadata handler

Restore control buttons and layout for Gallery tab with metadata textbox

Define _run_with_meta helper for global gallery handling

Show generation info only in global Gallery tab and remove per-tab metadata boxes

Merge branch 'main' of https://huggingface.co/spaces/R1000/ImageGen

Add select handler to display generation info for chosen gallery image

Update ui/layout.py

Revert repository to commit 1d4eb4e522c6572d69f5d06b1965a5ff48b664be

Reposition metadata textbox next to gallery in all tabs

Fix indentation in run_handlers for metadata support

Merge remote changes, keep our modifications for metadata

Add metadata textbox to galleries and return generation info

Enhance Gallery UI: add clear, download zip, improve layout

Handle tuple output from workflow and ensure tensor extraction in _gpu_logic

Remove SaveImage node, adjust workflow executor, restore tensor-to-image saving and deduplication

Restore image saving from tensor, dedupe duplicates, simplify result handling

Fix SyntaxError: align finally with try

Deduplicate SaveImage outputs and hash-filter duplicate files}

Deduplicate gallery images by base name to avoid duplicate extensions

Restrict demo.load to avoid repeated gallery loading

Fix syntax and deduplicate gallery loader

Deduplicate images in load_gallery_images to prevent duplicate thumbnails

Remove duplicate load_gallery_images definition to fix UnboundLocalError

Add Refresh button to Gallery tab to manually reload thumbnails

Update all result galleries to use filepath type for proper image handling

Update result gallery to use filepath type for correct handling

Separate Gallery tab to show thumbnails from output folder only, not duplicate result images

Fix run handler to duplicate output when gallery tab present

Set result gallery to 2 columns and height matching parameter blocks

Revert "Revert "Revert "Skip re-cloning ComfyUI on restart to speed up startup"""

Revert "Revert "Skip re-cloning ComfyUI on restart to speed up startup""

Revert "Skip re-cloning ComfyUI on restart to speed up startup"

Skip re-cloning ComfyUI on restart to speed up startup

Update run handlers to also update Gallery tab after generation

Adjust txt2img result gallery to match other tabs (columns=1, height=505)

Separate result gallery into right column and split parameter blocks as requested

Split Txt2Img base parameters into separate blocks as requested

Match original Txt2Img layout to Tinchote version: proper run column and direct prompt fields

Align Txt2Img layout with Img2Img: wrap base params and gallery in Row/Column

Adjust Txt2Img layout: wrap prompts in Row/Column and restore result gallery

Improve Txt2Img layout: use Row for model/run and align components like other tabs