- Categories:
- Cartoon
- Striptease
- Social Media Celebrities
- Celebrities:
-
Natalie Roush
Natalie Roush5123386
- Tags:
- Deepnude
Did the back/side by hand. Worked out okay.
Natalie Roush (born: April 9, 1995 [age 25]) is an American fitness/swimwear model. She quickly gained popularity on Instagram with over 500k followers. Her Youtube is where she is known for her fashion hauls and vlogging accumulating over 100k subscribers. Her TikTok profile has 1.2M Likes. 260.7K Fans. She has a Patreon you should all check out.
This video is made for entertainment purposes. No harm or ill-fate is intended.
This video took 20-30 hours. Broke video into frames and stitched it all back together after going in and doing touch-ups. Like classic animation.
Natalie Roush (born: April 9, 1995 [age 25]) is an American fitness/swimwear model. She quickly gained popularity on Instagram with over 500k followers. Her Youtube is where she is known for her fashion hauls and vlogging accumulating over 100k subscribers. Her TikTok profile has 1.2M Likes. 260.7K Fans. She has a Patreon you should all check out.
This video is made for entertainment purposes. No harm or ill-fate is intended.
This video took 20-30 hours. Broke video into frames and stitched it all back together after going in and doing touch-ups. Like classic animation.
Some of the time I will pre-crop. It is set to 512 x 512 so I work around it.--- I will sometimes crop closer in then mask and layer. My computing knowledge is low. I am patient. I have a love for illustration/animation. The 360 turn was by hand and not using the new
Theoretically, plug that into a small gui which remembers previous settings (namely cropping position) with a quick modification interface, and that should save some of the tedium.
Alternatively, if we limit the objective to just cropping the source, you could use openshot or your video editor of choice to do it (openshot specifically has a few curves for parameter change, so you could get a zoom in effect for example), export to images, run the model in batch, manually review the frames and correct / discard any bad ones, and stitch it back using the original project settings applied as an inverted mask (that is, make the source into a background).
By the way, did you add an automatic resizing of the input & output images, or does it still come out as 512 x 512?