gen_compare_v11 initial

gen_compare_v11/README.md

@@ -0,0 +1,102 @@
+# ControlNet Generation Comparison
+
+## control_v11p_sd15_canny
+
+|seed|conditional image|ref|diffusers|
+|:---:|:---:|:---:|:---:|
+|0|<a href="././control_images/converted//canny_vermeer_512x512.png"><img width="128" src="././control_images/converted//canny_vermeer_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_canny_0_0.png"><img width="128" src="./output_images/ref/output_p_sd15_canny_0_0.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_canny_0_0.png"><img width="128" src="./output_images/diffusers/output_p_sd15_canny_0_0.png"/></a>|
+|1|<a href="././control_images/converted//canny_vermeer_512x512.png"><img width="128" src="././control_images/converted//canny_vermeer_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_canny_0_1.png"><img width="128" src="./output_images/ref/output_p_sd15_canny_0_1.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_canny_0_1.png"><img width="128" src="./output_images/diffusers/output_p_sd15_canny_0_1.png"/></a>|
+|2|<a href="././control_images/converted//canny_vermeer_512x512.png"><img width="128" src="././control_images/converted//canny_vermeer_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_canny_0_2.png"><img width="128" src="./output_images/ref/output_p_sd15_canny_0_2.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_canny_0_2.png"><img width="128" src="./output_images/diffusers/output_p_sd15_canny_0_2.png"/></a>|
+|3|<a href="././control_images/converted//canny_vermeer_512x512.png"><img width="128" src="././control_images/converted//canny_vermeer_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_canny_0_3.png"><img width="128" src="./output_images/ref/output_p_sd15_canny_0_3.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_canny_0_3.png"><img width="128" src="./output_images/diffusers/output_p_sd15_canny_0_3.png"/></a>|
+## control_v11p_sd15_depth
+
+|seed|conditional image|ref|diffusers|
+|:---:|:---:|:---:|:---:|
+|0|<a href="././control_images/converted//zoe_pexels-sound-on-3760767_512x512.png"><img width="128" src="././control_images/converted//zoe_pexels-sound-on-3760767_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_depth_0_0.png"><img width="128" src="./output_images/ref/output_p_sd15_depth_0_0.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_depth_0_0.png"><img width="128" src="./output_images/diffusers/output_p_sd15_depth_0_0.png"/></a>|
+|1|<a href="././control_images/converted//zoe_pexels-sound-on-3760767_512x512.png"><img width="128" src="././control_images/converted//zoe_pexels-sound-on-3760767_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_depth_0_1.png"><img width="128" src="./output_images/ref/output_p_sd15_depth_0_1.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_depth_0_1.png"><img width="128" src="./output_images/diffusers/output_p_sd15_depth_0_1.png"/></a>|
+|2|<a href="././control_images/converted//zoe_pexels-sound-on-3760767_512x512.png"><img width="128" src="././control_images/converted//zoe_pexels-sound-on-3760767_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_depth_0_2.png"><img width="128" src="./output_images/ref/output_p_sd15_depth_0_2.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_depth_0_2.png"><img width="128" src="./output_images/diffusers/output_p_sd15_depth_0_2.png"/></a>|
+|3|<a href="././control_images/converted//zoe_pexels-sound-on-3760767_512x512.png"><img width="128" src="././control_images/converted//zoe_pexels-sound-on-3760767_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_depth_0_3.png"><img width="128" src="./output_images/ref/output_p_sd15_depth_0_3.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_depth_0_3.png"><img width="128" src="./output_images/diffusers/output_p_sd15_depth_0_3.png"/></a>|
+|0|<a href="././control_images/converted//midas_pexels-sound-on-3760767_512x512.png"><img width="128" src="././control_images/converted//midas_pexels-sound-on-3760767_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_depth_1_0.png"><img width="128" src="./output_images/ref/output_p_sd15_depth_1_0.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_depth_1_0.png"><img width="128" src="./output_images/diffusers/output_p_sd15_depth_1_0.png"/></a>|
+|1|<a href="././control_images/converted//midas_pexels-sound-on-3760767_512x512.png"><img width="128" src="././control_images/converted//midas_pexels-sound-on-3760767_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_depth_1_1.png"><img width="128" src="./output_images/ref/output_p_sd15_depth_1_1.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_depth_1_1.png"><img width="128" src="./output_images/diffusers/output_p_sd15_depth_1_1.png"/></a>|
+|2|<a href="././control_images/converted//midas_pexels-sound-on-3760767_512x512.png"><img width="128" src="././control_images/converted//midas_pexels-sound-on-3760767_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_depth_1_2.png"><img width="128" src="./output_images/ref/output_p_sd15_depth_1_2.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_depth_1_2.png"><img width="128" src="./output_images/diffusers/output_p_sd15_depth_1_2.png"/></a>|
+|3|<a href="././control_images/converted//midas_pexels-sound-on-3760767_512x512.png"><img width="128" src="././control_images/converted//midas_pexels-sound-on-3760767_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_depth_1_3.png"><img width="128" src="./output_images/ref/output_p_sd15_depth_1_3.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_depth_1_3.png"><img width="128" src="./output_images/diffusers/output_p_sd15_depth_1_3.png"/></a>|
+## control_v11p_sd15_mlsd
+
+|seed|conditional image|ref|diffusers|
+|:---:|:---:|:---:|:---:|
+|0|<a href="././control_images/converted//mlsd_room_512x512.png"><img width="128" src="././control_images/converted//mlsd_room_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_mlsd_0_0.png"><img width="128" src="./output_images/ref/output_p_sd15_mlsd_0_0.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_mlsd_0_0.png"><img width="128" src="./output_images/diffusers/output_p_sd15_mlsd_0_0.png"/></a>|
+|1|<a href="././control_images/converted//mlsd_room_512x512.png"><img width="128" src="././control_images/converted//mlsd_room_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_mlsd_0_1.png"><img width="128" src="./output_images/ref/output_p_sd15_mlsd_0_1.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_mlsd_0_1.png"><img width="128" src="./output_images/diffusers/output_p_sd15_mlsd_0_1.png"/></a>|
+|2|<a href="././control_images/converted//mlsd_room_512x512.png"><img width="128" src="././control_images/converted//mlsd_room_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_mlsd_0_2.png"><img width="128" src="./output_images/ref/output_p_sd15_mlsd_0_2.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_mlsd_0_2.png"><img width="128" src="./output_images/diffusers/output_p_sd15_mlsd_0_2.png"/></a>|
+|3|<a href="././control_images/converted//mlsd_room_512x512.png"><img width="128" src="././control_images/converted//mlsd_room_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_mlsd_0_3.png"><img width="128" src="./output_images/ref/output_p_sd15_mlsd_0_3.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_mlsd_0_3.png"><img width="128" src="./output_images/diffusers/output_p_sd15_mlsd_0_3.png"/></a>|
+## control_v11p_sd15_normalbae
+
+|seed|conditional image|ref|diffusers|
+|:---:|:---:|:---:|:---:|
+|0|<a href="././control_images/converted//normalbae_person_leaves_512x512.png"><img width="128" src="././control_images/converted//normalbae_person_leaves_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_normalbae_0_0.png"><img width="128" src="./output_images/ref/output_p_sd15_normalbae_0_0.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_normalbae_0_0.png"><img width="128" src="./output_images/diffusers/output_p_sd15_normalbae_0_0.png"/></a>|
+|1|<a href="././control_images/converted//normalbae_person_leaves_512x512.png"><img width="128" src="././control_images/converted//normalbae_person_leaves_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_normalbae_0_1.png"><img width="128" src="./output_images/ref/output_p_sd15_normalbae_0_1.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_normalbae_0_1.png"><img width="128" src="./output_images/diffusers/output_p_sd15_normalbae_0_1.png"/></a>|
+|2|<a href="././control_images/converted//normalbae_person_leaves_512x512.png"><img width="128" src="././control_images/converted//normalbae_person_leaves_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_normalbae_0_2.png"><img width="128" src="./output_images/ref/output_p_sd15_normalbae_0_2.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_normalbae_0_2.png"><img width="128" src="./output_images/diffusers/output_p_sd15_normalbae_0_2.png"/></a>|
+|3|<a href="././control_images/converted//normalbae_person_leaves_512x512.png"><img width="128" src="././control_images/converted//normalbae_person_leaves_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_normalbae_0_3.png"><img width="128" src="./output_images/ref/output_p_sd15_normalbae_0_3.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_normalbae_0_3.png"><img width="128" src="./output_images/diffusers/output_p_sd15_normalbae_0_3.png"/></a>|
+## control_v11p_sd15_openpose
+
+|seed|conditional image|ref|diffusers|
+|:---:|:---:|:---:|:---:|
+|0|<a href="././control_images/converted//openpose_pexels-sound-on-3760847_512x512.png"><img width="128" src="././control_images/converted//openpose_pexels-sound-on-3760847_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_openpose_0_0.png"><img width="128" src="./output_images/ref/output_p_sd15_openpose_0_0.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_openpose_0_0.png"><img width="128" src="./output_images/diffusers/output_p_sd15_openpose_0_0.png"/></a>|
+|1|<a href="././control_images/converted//openpose_pexels-sound-on-3760847_512x512.png"><img width="128" src="././control_images/converted//openpose_pexels-sound-on-3760847_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_openpose_0_1.png"><img width="128" src="./output_images/ref/output_p_sd15_openpose_0_1.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_openpose_0_1.png"><img width="128" src="./output_images/diffusers/output_p_sd15_openpose_0_1.png"/></a>|
+|2|<a href="././control_images/converted//openpose_pexels-sound-on-3760847_512x512.png"><img width="128" src="././control_images/converted//openpose_pexels-sound-on-3760847_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_openpose_0_2.png"><img width="128" src="./output_images/ref/output_p_sd15_openpose_0_2.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_openpose_0_2.png"><img width="128" src="./output_images/diffusers/output_p_sd15_openpose_0_2.png"/></a>|
+|3|<a href="././control_images/converted//openpose_pexels-sound-on-3760847_512x512.png"><img width="128" src="././control_images/converted//openpose_pexels-sound-on-3760847_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_openpose_0_3.png"><img width="128" src="./output_images/ref/output_p_sd15_openpose_0_3.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_openpose_0_3.png"><img width="128" src="./output_images/diffusers/output_p_sd15_openpose_0_3.png"/></a>|
+## control_v11p_sd15_scribble
+
+|seed|conditional image|ref|diffusers|
+|:---:|:---:|:---:|:---:|
+|0|<a href="././control_images/converted//hed_pexels-wallace-chuck-2383750_512x512.png"><img width="128" src="././control_images/converted//hed_pexels-wallace-chuck-2383750_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_scribble_0_0.png"><img width="128" src="./output_images/ref/output_p_sd15_scribble_0_0.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_scribble_0_0.png"><img width="128" src="./output_images/diffusers/output_p_sd15_scribble_0_0.png"/></a>|
+|1|<a href="././control_images/converted//hed_pexels-wallace-chuck-2383750_512x512.png"><img width="128" src="././control_images/converted//hed_pexels-wallace-chuck-2383750_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_scribble_0_1.png"><img width="128" src="./output_images/ref/output_p_sd15_scribble_0_1.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_scribble_0_1.png"><img width="128" src="./output_images/diffusers/output_p_sd15_scribble_0_1.png"/></a>|
+|2|<a href="././control_images/converted//hed_pexels-wallace-chuck-2383750_512x512.png"><img width="128" src="././control_images/converted//hed_pexels-wallace-chuck-2383750_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_scribble_0_2.png"><img width="128" src="./output_images/ref/output_p_sd15_scribble_0_2.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_scribble_0_2.png"><img width="128" src="./output_images/diffusers/output_p_sd15_scribble_0_2.png"/></a>|
+|3|<a href="././control_images/converted//hed_pexels-wallace-chuck-2383750_512x512.png"><img width="128" src="././control_images/converted//hed_pexels-wallace-chuck-2383750_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_scribble_0_3.png"><img width="128" src="./output_images/ref/output_p_sd15_scribble_0_3.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_scribble_0_3.png"><img width="128" src="./output_images/diffusers/output_p_sd15_scribble_0_3.png"/></a>|
+|0|<a href="././control_images/converted//pidi_pexels-sound-on-3760767_512x512.png"><img width="128" src="././control_images/converted//pidi_pexels-sound-on-3760767_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_scribble_1_0.png"><img width="128" src="./output_images/ref/output_p_sd15_scribble_1_0.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_scribble_1_0.png"><img width="128" src="./output_images/diffusers/output_p_sd15_scribble_1_0.png"/></a>|
+|1|<a href="././control_images/converted//pidi_pexels-sound-on-3760767_512x512.png"><img width="128" src="././control_images/converted//pidi_pexels-sound-on-3760767_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_scribble_1_1.png"><img width="128" src="./output_images/ref/output_p_sd15_scribble_1_1.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_scribble_1_1.png"><img width="128" src="./output_images/diffusers/output_p_sd15_scribble_1_1.png"/></a>|
+|2|<a href="././control_images/converted//pidi_pexels-sound-on-3760767_512x512.png"><img width="128" src="././control_images/converted//pidi_pexels-sound-on-3760767_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_scribble_1_2.png"><img width="128" src="./output_images/ref/output_p_sd15_scribble_1_2.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_scribble_1_2.png"><img width="128" src="./output_images/diffusers/output_p_sd15_scribble_1_2.png"/></a>|
+|3|<a href="././control_images/converted//pidi_pexels-sound-on-3760767_512x512.png"><img width="128" src="././control_images/converted//pidi_pexels-sound-on-3760767_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_scribble_1_3.png"><img width="128" src="./output_images/ref/output_p_sd15_scribble_1_3.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_scribble_1_3.png"><img width="128" src="./output_images/diffusers/output_p_sd15_scribble_1_3.png"/></a>|
+## control_v11p_sd15_seg
+
+|seed|conditional image|ref|diffusers|
+|:---:|:---:|:---:|:---:|
+|0|<a href="././control_images/converted//oneformer_adk20_house_512x512.png"><img width="128" src="././control_images/converted//oneformer_adk20_house_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_seg_0_0.png"><img width="128" src="./output_images/ref/output_p_sd15_seg_0_0.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_seg_0_0.png"><img width="128" src="./output_images/diffusers/output_p_sd15_seg_0_0.png"/></a>|
+|1|<a href="././control_images/converted//oneformer_adk20_house_512x512.png"><img width="128" src="././control_images/converted//oneformer_adk20_house_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_seg_0_1.png"><img width="128" src="./output_images/ref/output_p_sd15_seg_0_1.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_seg_0_1.png"><img width="128" src="./output_images/diffusers/output_p_sd15_seg_0_1.png"/></a>|
+|2|<a href="././control_images/converted//oneformer_adk20_house_512x512.png"><img width="128" src="././control_images/converted//oneformer_adk20_house_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_seg_0_2.png"><img width="128" src="./output_images/ref/output_p_sd15_seg_0_2.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_seg_0_2.png"><img width="128" src="./output_images/diffusers/output_p_sd15_seg_0_2.png"/></a>|
+|3|<a href="././control_images/converted//oneformer_adk20_house_512x512.png"><img width="128" src="././control_images/converted//oneformer_adk20_house_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_seg_0_3.png"><img width="128" src="./output_images/ref/output_p_sd15_seg_0_3.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_seg_0_3.png"><img width="128" src="./output_images/diffusers/output_p_sd15_seg_0_3.png"/></a>|
+|0|<a href="././control_images/converted//oneformer_coco_house_512x512.png"><img width="128" src="././control_images/converted//oneformer_coco_house_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_seg_1_0.png"><img width="128" src="./output_images/ref/output_p_sd15_seg_1_0.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_seg_1_0.png"><img width="128" src="./output_images/diffusers/output_p_sd15_seg_1_0.png"/></a>|
+|1|<a href="././control_images/converted//oneformer_coco_house_512x512.png"><img width="128" src="././control_images/converted//oneformer_coco_house_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_seg_1_1.png"><img width="128" src="./output_images/ref/output_p_sd15_seg_1_1.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_seg_1_1.png"><img width="128" src="./output_images/diffusers/output_p_sd15_seg_1_1.png"/></a>|
+|2|<a href="././control_images/converted//oneformer_coco_house_512x512.png"><img width="128" src="././control_images/converted//oneformer_coco_house_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_seg_1_2.png"><img width="128" src="./output_images/ref/output_p_sd15_seg_1_2.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_seg_1_2.png"><img width="128" src="./output_images/diffusers/output_p_sd15_seg_1_2.png"/></a>|
+|3|<a href="././control_images/converted//oneformer_coco_house_512x512.png"><img width="128" src="././control_images/converted//oneformer_coco_house_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_seg_1_3.png"><img width="128" src="./output_images/ref/output_p_sd15_seg_1_3.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_seg_1_3.png"><img width="128" src="./output_images/diffusers/output_p_sd15_seg_1_3.png"/></a>|
+|0|<a href="././control_images/converted//uniformer_house_512x512.png"><img width="128" src="././control_images/converted//uniformer_house_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_seg_2_0.png"><img width="128" src="./output_images/ref/output_p_sd15_seg_2_0.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_seg_2_0.png"><img width="128" src="./output_images/diffusers/output_p_sd15_seg_2_0.png"/></a>|
+|1|<a href="././control_images/converted//uniformer_house_512x512.png"><img width="128" src="././control_images/converted//uniformer_house_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_seg_2_1.png"><img width="128" src="./output_images/ref/output_p_sd15_seg_2_1.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_seg_2_1.png"><img width="128" src="./output_images/diffusers/output_p_sd15_seg_2_1.png"/></a>|
+|2|<a href="././control_images/converted//uniformer_house_512x512.png"><img width="128" src="././control_images/converted//uniformer_house_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_seg_2_2.png"><img width="128" src="./output_images/ref/output_p_sd15_seg_2_2.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_seg_2_2.png"><img width="128" src="./output_images/diffusers/output_p_sd15_seg_2_2.png"/></a>|
+|3|<a href="././control_images/converted//uniformer_house_512x512.png"><img width="128" src="././control_images/converted//uniformer_house_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_seg_2_3.png"><img width="128" src="./output_images/ref/output_p_sd15_seg_2_3.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_seg_2_3.png"><img width="128" src="./output_images/diffusers/output_p_sd15_seg_2_3.png"/></a>|
+## control_v11p_sd15_softedge
+
+|seed|conditional image|ref|diffusers|
+|:---:|:---:|:---:|:---:|
+|0|<a href="././control_images/converted//hed_pexels-wallace-chuck-2383750_512x512.png"><img width="128" src="././control_images/converted//hed_pexels-wallace-chuck-2383750_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_softedge_0_0.png"><img width="128" src="./output_images/ref/output_p_sd15_softedge_0_0.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_softedge_0_0.png"><img width="128" src="./output_images/diffusers/output_p_sd15_softedge_0_0.png"/></a>|
+|1|<a href="././control_images/converted//hed_pexels-wallace-chuck-2383750_512x512.png"><img width="128" src="././control_images/converted//hed_pexels-wallace-chuck-2383750_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_softedge_0_1.png"><img width="128" src="./output_images/ref/output_p_sd15_softedge_0_1.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_softedge_0_1.png"><img width="128" src="./output_images/diffusers/output_p_sd15_softedge_0_1.png"/></a>|
+|2|<a href="././control_images/converted//hed_pexels-wallace-chuck-2383750_512x512.png"><img width="128" src="././control_images/converted//hed_pexels-wallace-chuck-2383750_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_softedge_0_2.png"><img width="128" src="./output_images/ref/output_p_sd15_softedge_0_2.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_softedge_0_2.png"><img width="128" src="./output_images/diffusers/output_p_sd15_softedge_0_2.png"/></a>|
+|3|<a href="././control_images/converted//hed_pexels-wallace-chuck-2383750_512x512.png"><img width="128" src="././control_images/converted//hed_pexels-wallace-chuck-2383750_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_softedge_0_3.png"><img width="128" src="./output_images/ref/output_p_sd15_softedge_0_3.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_softedge_0_3.png"><img width="128" src="./output_images/diffusers/output_p_sd15_softedge_0_3.png"/></a>|
+|0|<a href="././control_images/converted//pidi_pexels-sound-on-3760767_512x512.png"><img width="128" src="././control_images/converted//pidi_pexels-sound-on-3760767_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_softedge_1_0.png"><img width="128" src="./output_images/ref/output_p_sd15_softedge_1_0.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_softedge_1_0.png"><img width="128" src="./output_images/diffusers/output_p_sd15_softedge_1_0.png"/></a>|
+|1|<a href="././control_images/converted//pidi_pexels-sound-on-3760767_512x512.png"><img width="128" src="././control_images/converted//pidi_pexels-sound-on-3760767_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_softedge_1_1.png"><img width="128" src="./output_images/ref/output_p_sd15_softedge_1_1.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_softedge_1_1.png"><img width="128" src="./output_images/diffusers/output_p_sd15_softedge_1_1.png"/></a>|
+|2|<a href="././control_images/converted//pidi_pexels-sound-on-3760767_512x512.png"><img width="128" src="././control_images/converted//pidi_pexels-sound-on-3760767_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_softedge_1_2.png"><img width="128" src="./output_images/ref/output_p_sd15_softedge_1_2.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_softedge_1_2.png"><img width="128" src="./output_images/diffusers/output_p_sd15_softedge_1_2.png"/></a>|
+|3|<a href="././control_images/converted//pidi_pexels-sound-on-3760767_512x512.png"><img width="128" src="././control_images/converted//pidi_pexels-sound-on-3760767_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_softedge_1_3.png"><img width="128" src="./output_images/ref/output_p_sd15_softedge_1_3.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_softedge_1_3.png"><img width="128" src="./output_images/diffusers/output_p_sd15_softedge_1_3.png"/></a>|
+## control_v11p_sd15_lineart
+
+|seed|conditional image|ref|diffusers|
+|:---:|:---:|:---:|:---:|
+|0|<a href="././control_images/converted//lineart_wolf_512x512.png"><img width="128" src="././control_images/converted//lineart_wolf_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_lineart_0_0.png"><img width="128" src="./output_images/ref/output_p_sd15_lineart_0_0.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_lineart_0_0.png"><img width="128" src="./output_images/diffusers/output_p_sd15_lineart_0_0.png"/></a>|
+|1|<a href="././control_images/converted//lineart_wolf_512x512.png"><img width="128" src="././control_images/converted//lineart_wolf_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_lineart_0_1.png"><img width="128" src="./output_images/ref/output_p_sd15_lineart_0_1.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_lineart_0_1.png"><img width="128" src="./output_images/diffusers/output_p_sd15_lineart_0_1.png"/></a>|
+|2|<a href="././control_images/converted//lineart_wolf_512x512.png"><img width="128" src="././control_images/converted//lineart_wolf_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_lineart_0_2.png"><img width="128" src="./output_images/ref/output_p_sd15_lineart_0_2.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_lineart_0_2.png"><img width="128" src="./output_images/diffusers/output_p_sd15_lineart_0_2.png"/></a>|
+|3|<a href="././control_images/converted//lineart_wolf_512x512.png"><img width="128" src="././control_images/converted//lineart_wolf_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15_lineart_0_3.png"><img width="128" src="./output_images/ref/output_p_sd15_lineart_0_3.png"/></a>|<a href="./output_images/diffusers/output_p_sd15_lineart_0_3.png"><img width="128" src="./output_images/diffusers/output_p_sd15_lineart_0_3.png"/></a>|
+## control_v11p_sd15s2_lineart_anime
+
+|seed|conditional image|ref|diffusers|
+|:---:|:---:|:---:|:---:|
+|0|<a href="././control_images/converted//lineartanime_anime2_512x512.png"><img width="128" src="././control_images/converted//lineartanime_anime2_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15s2_lineart_anime_0_0.png"><img width="128" src="./output_images/ref/output_p_sd15s2_lineart_anime_0_0.png"/></a>|<a href="./output_images/diffusers/output_p_sd15s2_lineart_anime_0_0.png"><img width="128" src="./output_images/diffusers/output_p_sd15s2_lineart_anime_0_0.png"/></a>|
+|1|<a href="././control_images/converted//lineartanime_anime2_512x512.png"><img width="128" src="././control_images/converted//lineartanime_anime2_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15s2_lineart_anime_0_1.png"><img width="128" src="./output_images/ref/output_p_sd15s2_lineart_anime_0_1.png"/></a>|<a href="./output_images/diffusers/output_p_sd15s2_lineart_anime_0_1.png"><img width="128" src="./output_images/diffusers/output_p_sd15s2_lineart_anime_0_1.png"/></a>|
+|2|<a href="././control_images/converted//lineartanime_anime2_512x512.png"><img width="128" src="././control_images/converted//lineartanime_anime2_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15s2_lineart_anime_0_2.png"><img width="128" src="./output_images/ref/output_p_sd15s2_lineart_anime_0_2.png"/></a>|<a href="./output_images/diffusers/output_p_sd15s2_lineart_anime_0_2.png"><img width="128" src="./output_images/diffusers/output_p_sd15s2_lineart_anime_0_2.png"/></a>|
+|3|<a href="././control_images/converted//lineartanime_anime2_512x512.png"><img width="128" src="././control_images/converted//lineartanime_anime2_512x512.png"/></a>|<a href="./output_images/ref/output_p_sd15s2_lineart_anime_0_3.png"><img width="128" src="./output_images/ref/output_p_sd15s2_lineart_anime_0_3.png"/></a>|<a href="./output_images/diffusers/output_p_sd15s2_lineart_anime_0_3.png"><img width="128" src="./output_images/diffusers/output_p_sd15s2_lineart_anime_0_3.png"/></a>|

gen_compare_v11/control_images.yaml

@@ -0,0 +1,34 @@
+p_sd15_canny:
+  - canny_vermeer_512x512.png
+p_sd15_depth:
+  - zoe_pexels-sound-on-3760767_512x512.png
+  - midas_pexels-sound-on-3760767_512x512.png
+p_sd15_lineart:
+  - lineart_wolf_512x512.png
+p_sd15_mlsd:
+  - mlsd_room_512x512.png
+p_sd15_normalbae:
+  - normalbae_person_leaves_512x512.png
+p_sd15_openpose:
+  - openpose_pexels-sound-on-3760847_512x512.png
+  # - openpose_pexels-sound-on-3760693_512x512.png
+  # - openpose_pexels-sound-on-3760767_512x512.png
+  # - openpose_pexels-wallace-chuck-2383750_512x512.png
+p_sd15_scribble:
+  - hed_pexels-wallace-chuck-2383750_512x512.png
+  - pidi_pexels-sound-on-3760767_512x512.png
+p_sd15_softedge:
+  - hed_pexels-wallace-chuck-2383750_512x512.png
+  - pidi_pexels-sound-on-3760767_512x512.png
+p_sd15_seg:
+  - oneformer_adk20_house_512x512.png
+  - oneformer_coco_house_512x512.png
+  - uniformer_house_512x512.png
+p_sd15s2_lineart_anime:
+  - lineartanime_anime2_512x512.png
+# p_sd15_inpaint:
+# e_sd15_shuffle:
+#   - colorshuffle_vermeer_512x512.png
+#   - shuffle_vermeer_512x512.png
+# e_sd15_ip2p:
+# u_sd15_tile:

gen_compare_v11/gen_diffusers_image.py

@@ -0,0 +1,73 @@
+# Diffusers' ControlNet Implementation Subjective Evaluation
+
+import einops
+import numpy as np
+import torch
+import sys
+import os
+import yaml
+
+from diffusers import StableDiffusionControlNetPipeline, ControlNetModel, DDIMScheduler
+
+from PIL import Image
+
+test_prompt = "best quality, extremely detailed"
+test_negative_prompt = "lowres, bad anatomy, worst quality, low quality"
+
+def generate_image(seed, prompt, negative_prompt, control, guess_mode=False):
+    latent = torch.randn((1,4,64,64), device="cpu", generator=torch.Generator(device="cpu").manual_seed(seed)).cuda()
+    image = pipe(
+        prompt=prompt,
+        negative_prompt=negative_prompt,
+        guidance_scale=4.0 if guess_mode else 9.0,
+        num_inference_steps=50 if guess_mode else 20,
+        latents=latent,
+        image=control,
+        # guess_mode=guess_mode,
+    ).images[0]
+    return image
+
+def control_images(control_image_folder, model_name):
+    with open('./control_images.yaml', 'r') as f:
+        d = yaml.safe_load(f)
+    filenames = d[model_name]
+    return [Image.open(f'{control_image_folder}/{fn}').convert("RGB") for fn in filenames]
+
+if __name__ == '__main__':
+    model_name = sys.argv[1]
+    control_image_folder = './control_images/converted/'
+    output_image_folder = './output_images/diffusers/'
+    os.makedirs(output_image_folder, exist_ok=True)
+
+    model_id = 'takuma104/control_v11'
+    subfolder = f'control_v11{model_name}'
+
+    controlnet = ControlNetModel.from_pretrained(model_id, subfolder=subfolder)
+    if model_name == 'p_sd15s2_lineart_anime':
+        base_model_id = 'Linaqruf/anything-v3.0'
+        base_model_revision = None
+    else:   
+        base_model_id = "runwayml/stable-diffusion-v1-5"
+        base_model_revision = 'non-ema'
+
+    pipe = StableDiffusionControlNetPipeline.from_pretrained(base_model_id, 
+                                                             revision=base_model_revision,
+                                                             controlnet=controlnet, 
+                                                             safety_checker=None).to("cuda")
+    pipe.scheduler = DDIMScheduler.from_config(pipe.scheduler.config)
+
+    for i, control_image in enumerate(control_images(control_image_folder, model_name)):
+        control = np.array(control_image)[:,:,::-1].copy()
+        control = torch.from_numpy(control).float().cuda() / 255.0
+        control = torch.stack([control for _ in range(1)], dim=0)
+        control = einops.rearrange(control, 'b h w c -> b c h w').clone()
+
+        # if model_name == 'p_sd15_normalbae': # workaround, this should not be necessary
+        #     control = torch.flip(control, dims=[1]) # RGB -> BGR
+
+        for seed in range(4):
+            image = generate_image(seed=seed, 
+                                   prompt=test_prompt, 
+                                   negative_prompt=test_negative_prompt, 
+                                   control=control)
+            image.save(f'{output_image_folder}output_{model_name}_{i}_{seed}.png')

gen_compare_v11/gen_diffusers_image.sh

@@ -0,0 +1,21 @@
+#!/bin/bash
+
+models=(
+    "p_sd15_canny" 
+    "p_sd15_depth" 
+    "p_sd15_mlsd" 
+    "p_sd15_normalbae" 
+    "p_sd15_openpose" 
+    "p_sd15_scribble" 
+    "p_sd15_seg" 
+    "p_sd15_softedge"
+    "p_sd15_lineart"
+    "p_sd15s2_lineart_anime"
+)
+# models=("p_sd15_inpaint" "e_sd15_shuffle" "e_sd15_ip2p" "u_sd15_tile")
+
+for model in "${models[@]}"
+do
+    echo $model
+    python gen_diffusers_image.py $model
+done

gen_compare_v11/gen_markdown.py

@@ -0,0 +1,66 @@
+import sys
+import yaml
+import os
+
+cnet_models = [
+    "p_sd15_canny",
+    "p_sd15_depth", 
+    "p_sd15_mlsd" ,
+    "p_sd15_normalbae", 
+    "p_sd15_openpose", 
+    "p_sd15_scribble", 
+    "p_sd15_seg", 
+    "p_sd15_softedge",
+    "p_sd15_lineart",
+    "p_sd15s2_lineart_anime",
+    # "p_sd15_inpaint",
+    # "e_sd15_shuffle",
+    # "e_sd15_ip2p",
+    # "u_sd15_tile",
+]
+
+processors = ["ref", "diffusers"]
+
+print(f"# ControlNet Generation Comparison")
+print("")
+
+def control_image_filenames(model_name):
+    with open('./control_images.yaml', 'r') as f:
+        d = yaml.safe_load(f)
+    return d[model_name]
+
+def print_header():
+    print("|seed|conditional image|", end="")
+    for proc in processors:
+        print(f"{proc}|", end="")
+    print("")
+
+    print("|", end="")
+    for _ in range(len(processors) + 2):
+        print(":---:|", end="")
+    print("")
+
+def get_url(path):
+    url = "./"
+    # url = 'https://huggingface.co/takuma104/controlnet_dev/resolve/main/gen_compare_v11/'
+    return url + path
+
+control_image_folder = './control_images/converted/'
+
+for cnet in cnet_models:
+    print(f"## control_v11{cnet}")
+    print("")
+    print_header()
+
+    for i, control_image_fn in enumerate(control_image_filenames(cnet)):
+        for seed in range(4):
+            print(f"|{seed}|", end="")
+
+            url = get_url(path=f"{control_image_folder}/{control_image_fn}")
+            print(f'<a href="{url}"><img width="128" src="{url}"/></a>|', end="")
+
+            for proc in processors:
+                url = get_url(path=f"output_images/{proc}/output_{cnet}_{i}_{seed:01d}.png")
+                print(f'<a href="{url}"><img width="128" src="{url}"/></a>|', end="")
+
+            print("")

gen_compare_v11/gen_markdowns.sh

@@ -0,0 +1,4 @@
+#!/bin/bash
+
+python gen_markdown.py > README.md
+

gen_compare_v11/gen_reference_image.py

@@ -0,0 +1,77 @@
+# from https://github.com/lllyasviel/ControlNet/blob/main/gradio_canny2image.py
+
+import einops
+import numpy as np
+import torch
+from PIL import Image
+import sys
+import os
+import yaml
+
+CONTROL_NET_PATH = '../../../ControlNet-v1-1-nightly'
+CONTROL_NET_MODEL_PATH = '../../ControlNet-v1-1'
+sys.path.append(CONTROL_NET_PATH)
+
+from share import *
+from pytorch_lightning import seed_everything
+from cldm.model import create_model, load_state_dict
+from cldm.ddim_hacked import DDIMSampler
+from diffusers.utils import load_image
+
+test_prompt = "best quality, extremely detailed"
+test_negative_prompt = "lowres, bad anatomy, worst quality, low quality"
+
+@torch.no_grad()
+def generate(prompt, n_prompt, seed, control, ddim_steps=20, eta=0.0, scale=9.0, H=512, W=512, strength = 1.0, guess_mode=False):
+    seed_everything(seed)
+
+    cond = {"c_concat": [control], "c_crossattn": [model.get_learned_conditioning([prompt] * num_samples)]}
+    un_cond = {"c_concat": None if guess_mode else [control], "c_crossattn": [model.get_learned_conditioning([n_prompt] * num_samples)]}
+    shape = (4, H // 8, W // 8)
+    
+    model.control_scales = [strength * (0.825 ** float(12 - i)) for i in range(13)] if guess_mode else ([strength] * 13)  # Magic number. IDK why. Perhaps because 0.825**12<0.01 but 0.826**12>0.01
+    latent = torch.randn((1,) + shape, device="cpu", generator=torch.Generator(device="cpu").manual_seed(seed)).cuda()
+    samples, intermediates = ddim_sampler.sample(ddim_steps, num_samples,
+                                                    shape, cond, x_T=latent,
+                                                    verbose=False, eta=eta,
+                                                    unconditional_guidance_scale=scale,
+                                                    unconditional_conditioning=un_cond)
+    x_samples = model.decode_first_stage(samples)
+    x_samples = (einops.rearrange(x_samples, 'b c h w -> b h w c') * 127.5 + 127.5).cpu().numpy().clip(0, 255).astype(np.uint8)
+ 
+    return Image.fromarray(x_samples[0])
+
+def control_images(control_image_folder, model_name):
+    with open('./control_images.yaml', 'r') as f:
+        d = yaml.safe_load(f)
+    filenames = d[model_name]
+    return [Image.open(f'{control_image_folder}/{fn}').convert("RGB") for fn in filenames]
+
+if __name__ == '__main__':
+    model_name = sys.argv[1]
+
+    control_image_folder = './control_images/converted/'
+    output_image_folder = './output_images/ref/'
+    os.makedirs(output_image_folder, exist_ok=True)
+
+    if model_name == 'p_sd15s2_lineart_anime':
+        base_model_file = 'anything-v3-full.safetensors'
+    else:
+        base_model_file = 'v1-5-pruned.ckpt'
+
+    num_samples = 1
+    model = create_model(f'{CONTROL_NET_MODEL_PATH}/control_v11{model_name}.yaml').cpu()
+    model.load_state_dict(load_state_dict(f'{CONTROL_NET_PATH}/models/{base_model_file}', location='cuda'), strict=False)
+    model.load_state_dict(load_state_dict(f'{CONTROL_NET_MODEL_PATH}/control_v11{model_name}.pth', location='cuda'), strict=False)
+    model = model.cuda()
+    ddim_sampler = DDIMSampler(model)
+
+    for i, control_image in enumerate(control_images(control_image_folder, model_name)):
+        control = np.array(control_image)[:,:,::-1].copy()
+        control = torch.from_numpy(control).float().cuda() / 255.0
+        control = torch.stack([control for _ in range(num_samples)], dim=0)
+        control = einops.rearrange(control, 'b h w c -> b c h w').clone()
+
+        for seed in range(4):
+            image = generate(test_prompt, test_negative_prompt, seed=seed, control=control)
+            image.save(f'{output_image_folder}output_{model_name}_{i}_{seed}.png')

gen_compare_v11/gen_reference_image.sh

@@ -0,0 +1,21 @@
+#!/bin/bash
+
+models=(
+    "p_sd15_canny" 
+    "p_sd15_depth" 
+    "p_sd15_mlsd" 
+    "p_sd15_normalbae" 
+    "p_sd15_openpose" 
+    "p_sd15_scribble" 
+    "p_sd15_seg" 
+    "p_sd15_softedge"
+    "p_sd15_lineart"
+    "p_sd15s2_lineart_anime"
+)
+# models=("p_sd15_inpaint" "e_sd15_shuffle" "e_sd15_ip2p" "u_sd15_tile")
+
+for model in "${models[@]}"
+do
+    echo $model
+    python gen_reference_image.py $model
+done