348 lines
11 KiB
Python
348 lines
11 KiB
Python
import math, time, os
|
|
import numpy as np
|
|
from PIL import Image
|
|
from modules.ui import plaintext_to_html
|
|
import modules.shared as shared
|
|
|
|
from .helpers import (
|
|
fix_env_Path_ffprobe,
|
|
closest_upper_divisible_by_eight,
|
|
load_model_from_setting,
|
|
do_upscaleImg,
|
|
predict_upscalesize
|
|
)
|
|
from .sd_helpers import renderImg2Img, renderTxt2Img
|
|
from .image import shrink_and_paste_on_blank
|
|
from .video import write_video
|
|
|
|
|
|
def create_zoom(
|
|
prompts_array,
|
|
negative_prompt,
|
|
num_outpainting_steps,
|
|
guidance_scale,
|
|
num_inference_steps,
|
|
custom_init_image,
|
|
custom_exit_image,
|
|
video_frame_rate,
|
|
video_zoom_mode,
|
|
video_start_frame_dupe_amount,
|
|
video_last_frame_dupe_amount,
|
|
inpainting_denoising_strength,
|
|
inpainting_mask_blur,
|
|
inpainting_fill_mode,
|
|
inpainting_full_res,
|
|
inpainting_padding,
|
|
zoom_speed,
|
|
seed,
|
|
outputsizeW,
|
|
outputsizeH,
|
|
batchcount,
|
|
sampler,
|
|
upscale_do,
|
|
upscaler_name,
|
|
upscale_by,
|
|
progress=None,
|
|
):
|
|
for i in range(batchcount):
|
|
print(f"Batch {i+1}/{batchcount}")
|
|
result = create_zoom_single(
|
|
prompts_array,
|
|
negative_prompt,
|
|
num_outpainting_steps,
|
|
guidance_scale,
|
|
num_inference_steps,
|
|
custom_init_image,
|
|
custom_exit_image,
|
|
video_frame_rate,
|
|
video_zoom_mode,
|
|
video_start_frame_dupe_amount,
|
|
video_last_frame_dupe_amount,
|
|
inpainting_denoising_strength,
|
|
inpainting_mask_blur,
|
|
inpainting_fill_mode,
|
|
inpainting_full_res,
|
|
inpainting_padding,
|
|
zoom_speed,
|
|
seed,
|
|
outputsizeW,
|
|
outputsizeH,
|
|
sampler,
|
|
upscale_do,
|
|
upscaler_name,
|
|
upscale_by,
|
|
progress,
|
|
)
|
|
return result
|
|
|
|
|
|
def create_zoom_single(
|
|
prompts_array,
|
|
negative_prompt,
|
|
num_outpainting_steps,
|
|
guidance_scale,
|
|
num_inference_steps,
|
|
custom_init_image,
|
|
custom_exit_image,
|
|
video_frame_rate,
|
|
video_zoom_mode,
|
|
video_start_frame_dupe_amount,
|
|
video_last_frame_dupe_amount,
|
|
inpainting_denoising_strength,
|
|
inpainting_mask_blur,
|
|
inpainting_fill_mode,
|
|
inpainting_full_res,
|
|
inpainting_padding,
|
|
zoom_speed,
|
|
seed,
|
|
outputsizeW,
|
|
outputsizeH,
|
|
sampler,
|
|
upscale_do,
|
|
upscaler_name,
|
|
upscale_by,
|
|
progress=None,
|
|
):
|
|
fix_env_Path_ffprobe()
|
|
|
|
prompts = {}
|
|
for x in prompts_array:
|
|
try:
|
|
key = int(x[0])
|
|
value = str(x[1])
|
|
prompts[key] = value
|
|
except ValueError:
|
|
pass
|
|
assert len(prompts_array) > 0, "prompts is empty"
|
|
|
|
width = closest_upper_divisible_by_eight(outputsizeW)
|
|
height = closest_upper_divisible_by_eight(outputsizeH)
|
|
|
|
# smart upscale: only keyframes to upscale,
|
|
# therefore change size as state for the hole process
|
|
if upscale_do:
|
|
widthInterp,heightInterp,void = predict_upscalesize(outputsizeW, outputsizeH,upscale_by)
|
|
else:
|
|
widthInterp = width
|
|
heightInterp = height
|
|
|
|
current_image = Image.new(mode="RGBA", size=(width, height))
|
|
mask_image = np.array(current_image)[:, :, 3]
|
|
mask_image = Image.fromarray(255 - mask_image).convert("RGB")
|
|
current_image = current_image.convert("RGB")
|
|
current_seed = seed
|
|
|
|
if custom_init_image:
|
|
if upscale_do:
|
|
print(f"using Custom Initial Image, upscaling using {upscaler_name}")
|
|
current_image = do_upscaleImg(custom_init_image,True,upscaler_name,(width,height))
|
|
|
|
else:
|
|
print(f"using Custom Initial Image, simple resizing to {width} x {height}")
|
|
current_image = custom_init_image.resize(
|
|
(width, height), resample=Image.LANCZOS
|
|
)
|
|
else:
|
|
load_model_from_setting(
|
|
"infzoom_txt2img_model", progress, "Loading Model for txt2img: "
|
|
)
|
|
|
|
processed, newseed = renderTxt2Img(
|
|
prompts[min(k for k in prompts.keys() if k >= 0)],
|
|
negative_prompt,
|
|
sampler,
|
|
num_inference_steps,
|
|
guidance_scale,
|
|
current_seed,
|
|
width,
|
|
height,
|
|
)
|
|
|
|
if(len(processed.images) > 0):
|
|
current_image = processed.images[0]
|
|
current_seed = newseed
|
|
|
|
|
|
num_interpol_frames = round(video_frame_rate * zoom_speed)
|
|
|
|
all_frames = []
|
|
|
|
if upscale_do:
|
|
if progress:
|
|
progress(0, desc="upscaling inital txt2image")
|
|
all_frames.append(do_upscaleImg(current_image, upscale_do, upscaler_name, upscale_by))
|
|
else:
|
|
all_frames.append(current_image)
|
|
### end of txt2img
|
|
|
|
## begin img2img
|
|
load_model_from_setting(
|
|
"infzoom_inpainting_model", progress, "Loading Model for inpainting/img2img: "
|
|
)
|
|
|
|
for i in range(num_outpainting_steps):
|
|
mask_width = math.trunc(outputsizeH / 4) # was initially 512px => 128px
|
|
mask_height = math.trunc(outputsizeH / 4) # was initially 512px => 128px
|
|
print_out = (
|
|
"Outpaint step: "
|
|
+ str(i + 1)
|
|
+ " / "
|
|
+ str(num_outpainting_steps)
|
|
+ " Seed: "
|
|
+ str(current_seed)
|
|
)
|
|
print(print_out)
|
|
if progress:
|
|
progress(((i + 1) / num_outpainting_steps), desc=print_out)
|
|
|
|
prev_image_fix = current_image
|
|
prev_image = shrink_and_paste_on_blank(current_image, mask_width, mask_height)
|
|
current_image = prev_image
|
|
|
|
# create mask (black image with white mask_width width edges)
|
|
mask_image = np.array(current_image)[:, :, 3]
|
|
mask_image = Image.fromarray(255 - mask_image).convert("RGB")
|
|
|
|
# inpainting step
|
|
current_image = current_image.convert("RGB")
|
|
|
|
if custom_exit_image and ((i + 1) == num_outpainting_steps):
|
|
if upscale_do:
|
|
if progress:
|
|
progress(0, desc=f"upscaling Custom Inital image using {upscaler_name}")
|
|
|
|
print (f"upscaling Custom Inital image using {upscaler_name}")
|
|
current_image = do_upscaleImg(custom_exit_image, upscale_do, upscaler_name, (outputsizeW, outputsizeH))
|
|
else:
|
|
print("using Custom Exit Image, simple resizing")
|
|
current_image = custom_exit_image.resize(
|
|
(outputsizeW, outputsizeH), resample=Image.LANCZOS
|
|
)
|
|
|
|
else:
|
|
processed, newseed = renderImg2Img(
|
|
prompts[max(k for k in prompts.keys() if k <= i)],
|
|
negative_prompt,
|
|
sampler,
|
|
num_inference_steps,
|
|
guidance_scale,
|
|
current_seed,
|
|
outputsizeW,
|
|
outputsizeH,
|
|
current_image,
|
|
mask_image,
|
|
inpainting_denoising_strength,
|
|
inpainting_mask_blur,
|
|
inpainting_fill_mode,
|
|
inpainting_full_res,
|
|
inpainting_padding,
|
|
)
|
|
|
|
if(len(processed.images) > 0):
|
|
current_image = processed.images[0]
|
|
current_seed = newseed
|
|
|
|
if(len(processed.images) > 0):
|
|
current_image.paste(prev_image, mask=prev_image)
|
|
### end img2img
|
|
|
|
### begin interpolation
|
|
# from here in case of upscale everything is XL:
|
|
if upscale_do:
|
|
if progress:
|
|
progress(0, desc="upscaling curr+prevImage")
|
|
|
|
current_imageXL = do_upscaleImg(current_image, upscale_do, upscaler_name, upscale_by)
|
|
prev_image_fixXL = do_upscaleImg(prev_image_fix, upscale_do, upscaler_name, upscale_by)
|
|
mask_widthXL = math.trunc(widthInterp / 4) # was initially 512px => 128px
|
|
mask_heightXL = math.trunc(heightInterp / 4) # was initially 512px => 128px
|
|
else:
|
|
current_imageXL = current_image
|
|
mask_widthXL = math.trunc(outputsizeW / 4) # was initially 512px => 128px
|
|
mask_heightXL = math.trunc(outputsizeH / 4) # was initially 512px => 128px
|
|
|
|
# interpolation steps between 2 inpainted (upscaled) images (=sequential zoom and crop)
|
|
for j in range(num_interpol_frames - 1):
|
|
interpol_image = current_imageXL
|
|
|
|
interpol_width = round(
|
|
(
|
|
1
|
|
- (1 - 2 * mask_widthXL / widthInterp)
|
|
** (1 - (j + 1) / num_interpol_frames)
|
|
)
|
|
* widthInterp
|
|
/ 2
|
|
)
|
|
|
|
interpol_height = round(
|
|
(
|
|
1
|
|
- (1 - 2 * mask_heightXL / heightInterp)
|
|
** (1 - (j + 1) / num_interpol_frames)
|
|
)
|
|
* heightInterp
|
|
/ 2
|
|
)
|
|
|
|
interpol_image = interpol_image.crop(
|
|
(
|
|
interpol_width,
|
|
interpol_height,
|
|
widthInterp - interpol_width,
|
|
heightInterp - interpol_height,
|
|
)
|
|
)
|
|
|
|
interpol_image = interpol_image.resize((widthInterp, heightInterp))
|
|
|
|
# paste the higher resolution previous image in the middle to avoid drop in quality caused by zooming
|
|
interpol_width2 = round(
|
|
(1 - (widthInterp - 2 * mask_widthXL) / (widthInterp - 2 * interpol_width))
|
|
/ 2
|
|
* widthInterp
|
|
)
|
|
|
|
interpol_height2 = round(
|
|
(1 - (heightInterp - 2 * mask_heightXL) / (heightInterp - 2 * interpol_height))
|
|
/ 2
|
|
* heightInterp
|
|
)
|
|
|
|
prev_image_fix_crop = shrink_and_paste_on_blank(
|
|
prev_image_fixXL, interpol_width2, interpol_height2
|
|
)
|
|
|
|
interpol_image.paste(prev_image_fix_crop, mask=prev_image_fix_crop)
|
|
|
|
all_frames.append(interpol_image)
|
|
|
|
all_frames.append(current_image)
|
|
|
|
video_file_name = "infinite_zoom_" + str(int(time.time())) + ".mp4"
|
|
output_path = shared.opts.data.get(
|
|
"infzoom_outpath", shared.opts.data.get("outdir_img2img_samples")
|
|
)
|
|
save_path = os.path.join(
|
|
output_path, shared.opts.data.get("infzoom_outSUBpath", "infinite-zooms")
|
|
)
|
|
if not os.path.exists(save_path):
|
|
os.makedirs(save_path)
|
|
out = os.path.join(save_path, video_file_name)
|
|
write_video(
|
|
out,
|
|
all_frames,
|
|
video_frame_rate,
|
|
video_zoom_mode,
|
|
int(video_start_frame_dupe_amount),
|
|
int(video_last_frame_dupe_amount),
|
|
)
|
|
|
|
return (
|
|
out,
|
|
processed.images,
|
|
processed.js(),
|
|
plaintext_to_html(processed.info),
|
|
plaintext_to_html(""),
|
|
)
|