AI image generation

anthropomorphic elk wearing boxing gloves, fighting stance, angry expression, drunk, minimal line drawing, outlines
elkoman.png

This one was a blend of two to make Todd Squidward. Aka, the look Todd gives you when he finds out you haven't purchased Starfield yet.
todd squidward.png

Anothet blend:
" rIsE anD sHIne, mIsTeR cOuRieR… wAke Up aND smElL tHe SarSaPariLLa"
Mister House and the G-Man
g-house.png
 
Columbo dressed as Sherlock Holmes but in 1970s New York

PYjq6VT.jpg

I'm seeing snippets/features from other actors, except Peter Falk

A little Clint Eastwood, a lot of Johnny Cash, some John Pertwee, and even a smattering of my cocaine

Idk maybe it's just me
 
A friend tipped me to Fooocus, yet another AI image generator you run on your own computer (running it was also dead simple, although I think it might be using the python stuff I already had installed from other programs - not sure about that. Just had to download one 6GB model). But this one focuses on simplicity and quality, taking tips from Midjourney.

Here's "Horror stoner metal landscape painting of General Tso's Compressor, a steampunk style siege engine"

2024-01-03_20-56-27_3608.png
I asked for some outpainting, as in "continue the image on all directions" so it could fit better on 125B.
2024-01-03_21-06-18_1790.png
...I guess technically that is what I asked for.

Here's a bonus General Tso picture from earlier on:
2024-01-03_20-27-38_6173.png
 
Fooocus is much faster to get good results with than the previous ones I've used.

Chalumeau
Chalumeau_2.png
Good Vibes
Good Vibes.png
Timbre Man (obviously Timber Man here)
Timberman.png
General Tso's compressor (in the middle, presumably - the tiny boat at the bottom will be outside the print area so no need to remove it manually, and I had to drop the quality a bit to get it to upload here so more compression artifacts than in the OG picture)
Tso_compressor_2.png

Tso took some more work, the others were just a couple of generations. I enabled the SAI Fantasy Art and SAI Comic Book styles in addition to the default ones and raised guidance scale to 8. Plus a bit of extra tweaking with outpainting and upscaling to get the pictures big enough.
 
A friend tipped me to Fooocus, yet another AI image generator you run on your own computer (running it was also dead simple, although I think it might be using the python stuff I already had installed from other programs - not sure about that. Just had to download one 6GB model). But this one focuses on simplicity and quality, taking tips from Midjourney.

I started to install this on my laptop last night, but quickly found out I'm running out of HD space. This SSD only has 256 gigs, and I need to get a bigger one soon.
 
I started to install this on my laptop last night, but quickly found out I'm running out of HD space. This SSD only has 256 gigs, and I need to get a bigger one soon.
Did it download the python stuff for you? Or was the 6GB model too much? I'm not quite sure what exactly it downloads.
 
Video is just a ton of images in a row, right?

I did a music video for my main project using AI:


Fooocus for the base images and then using ComfyUI to make videos out of those, Flowframes for frame interpolation (ComfyUI outputs like 28 frames or something like that) and Davinci Resolve to edit it together. I'd say there were like 3-4 bad videos for every good video, and the good ones were all very short and most were just panning or zooming or such. It's fine, but I'd say waiting for Sora is probably the best option at this point if you want to do a music video yourself.

(it's not actually played before an audience either, we just did a faster take of that one song for fun and I faked an audience in post - it's all lies!)
 
Video is just a ton of images in a row, right?

I did a music video for my main project using AI:


Fooocus for the base images and then using ComfyUI to make videos out of those, Flowframes for frame interpolation (ComfyUI outputs like 28 frames or something like that) and Davinci Resolve to edit it together. I'd say there were like 3-4 bad videos for every good video, and the good ones were all very short and most were just panning or zooming or such. It's fine, but I'd say waiting for Sora is probably the best option at this point if you want to do a music video yourself.

(it's not actually played before an audience either, we just did a faster take of that one song for fun and I faked an audience in post - it's all lies!)
i cannot wait for sora. we've used it to help design the interior of our brewery. a contractor asked who our designer was, it blew her mind when she found out it was AI.

and the art, oh the art! i've made some super silly stuff. lots of good stuff in here, well done guys.
 
i cannot wait for sora. we've used it to help design the interior of our brewery. a contractor asked who our designer was, it blew her mind when she found out it was AI.

and the art, oh the art! i've made some super silly stuff. lots of good stuff in here, well done guys.
Where's the brewery?
 
Back
Top