cgi Save Point: Concept Art

The following concept art comes directly from the current pipeline, but we haven't yet gotten it fully stable for animation. It's way closer than it was, but in the mean time, here are some frame grabs from the AI layers in progress.

00370-3640275578.png

00371-3640275579.png

00375-2897079489.png

00424-3106824358.png

00423-3106824357.png

00448-4182603204.png

00450-1692978742.png

00457-4037472810.png

00461-4037472814.png

00505-2576896614.png

00508-2874724660.png

00537-2874724689.png

00179-1533196975.png

00182-1533196978.png

00558-2874724710.png

00563-2272083760.png
 
As a language model, I cannot condone the use of the word "elegant"

I've just started saying this in real life now, the repetition got to me.
I've been paying for the advanced model and its a lot less frustrating.

I still use the old one for really simple stuff... but oh man, the old one is bad.
Today it told me that light takes 1.28 seconds to get to the moon, therefore it's impossible for anything to reach the moon in a minute.

And it kept insisting that it's logic was right, nonsense reply after nonsense reply.

I asked the 4.0 and I got a detailed awesome answer, including the g-forces when you pass out and the g-forces you die, and how to account for that, and how fast my wizards can actually get to the moon avoiding for a lethal dose of 50 g-forces. Pretty sweet.

Writing their moon adventure now.
 
I'm actually really looking forward to checking it out. I'm just knee deep in a huge que of tasks right now. I'm trying to throw together this Labyrinth sub product over the summer and get it out there earning money. I don't know, I might get around to doing a pure research day on it the next time I get highway hypnosis from editing cells. I'd love it if it could do some of the really time consuming grunt work for me, such as top down maze planning.

Can it output graphs, pictures, spreadsheet files, or other more useful stuff yet? I can write just fine, but it is a lot of labor to blueprint a city and draw a map, then rebuild it to have certain numerical properties. (such as making it so no one can run off with my prize money by just taking 30 left turns in a row)
 
Very interesting research Nate. Its all very consistent. Looking forward on how you gonna develop it further.
Thanks, It's a ton of work every day, but It's been kind of an amazing experience the last few months, just watching a whole new world of possibilities open up. I've actually developed it a lot further already, in case you missed the videos on that. I put out so many that I can't imagine any one person has the time to watch them all. Since it might be of general interest to people who stumble across this thread, here's a little collection of videos relevant to my progress bridging UE5 into a heavily modded stable diffusion. The video "Hybrid" is the one where the real breakthrough took place. There's thousands of hours of work to be done still on just the pipeline and art style, I'm not fully satisfied with it yet, but the system is now working, so I'm building the mini project "The Labyrinth" in it's final form now, in an effort to get it shipped out so I can start the actual story of Save Point, which I'm way more excited about.




 
Right now, there is a bottleneck in the pipeline, where I'm producing pretty high grade output, and capable of smooth animation, but this new module I'm using to automate animating scenes is way behind the rest of the pipeline. So here I'll show some images from the midstage of the pipeline, the stills, where it's the full quality I'm actually producing, and then later for comparison, the final output based on those directives and genetics. Unfortunately had to leave out some of the best pictures, because spoilers. Anyway, this is what the output looks like right now, before it gets degraded in the animation automator stage, which reduces resolution to about 300 x 700 for now, and needs to be replaced.

00009-1647134694.png


00034-3842252222.png



00418-1764126803.png


00425-1764126810.png


00480-4248998472.png


00509-842376135.png


00505-842376131.png


00528-466507567.png


00540-3890506328.png


00536-3890506324.png


00555-3890506343.png


00607-1699628388.png


00639-1051868478.png


00642-1051868481.png


00649-1051868488.png


00017-1051868506.png


00023-1051868512.png


00037-1051868526.png


00048-1857856221.png



00051-1857856224.png


00057-1857856230.png



00309-3566804270.png


00377-3566804338.png
 
those last two frames - did you figure out how to animate the fire?
Sort of, the solution could be better. Ultimately, I've solved everything on some level. The huge deal is that almost everything needs to be solved at full automation level. So I can make that fire animate photorealistic, if I stop and work on one scene for an hour. Can't do a branching film with 10s of thousands of scenes manually, so I have to have a tech end solution that works. One is coming though, so I'm just bulldozing ahead for now, and when the fixed module is available, I'll swap it in and rerun all the sequences.

So the answer is "Ish". I can animate everything at a bad level fast, a good level slow, and a great level very slow. Slow just isn't an option for this. I'm going to be publishing a video sometime in the next few days that will have some animated fire and smoke in it, at the current level. Of course you saw the Snapshot video, which had the current fire and smoke automated animation tech on display a few weeks back.

 
Closing in slowly on the original desired outcome. This last week we finished training the cine frame LORA (sub-brain AI). It's job is to simply take the output from the midstage, compare it to actual cinema frames, and give them more of the look of classic 35 mm film. I've got it maybe half working at this point.

Also, botanical animals and electronic tigers.



00228-87968587.png


00211-3259445219.png


00210-3259445218.png


00110-3347570174.png


00116-3347570180.png


00099-3830868440.png


00124-3347570188.png


00201-3259445209.png


00202-3259445210.png



00214-3259445222.png



00233-87968592.png


00046-433565914.png


00142-1804005875.png


00396-70242362.png


00112-3718883283.png


00111-3718883282.png


00180-1947654148.png



00310-2848862135.png



00429-1675762104.png



00432-1675762107.png



00446-1675762121.png


00075-2618456282.png


00088-2618456295.png


00101-1264197262.png


00087-530500446.png


00088-530500447.png


00115-484245160.png


00134-2945186499.png


00161-2945186526.png


00164-2945186529.png


00206-1465102386.png


00184-3311972898.png


00185-3311972899.png


00159-208636452.png


00261-837405322.png


00263-837405324.png


00334-837405395.png


00332-837405393.png


00308-3039821616.png
 

Attachments

  • 00215-1653137838.png
    00215-1653137838.png
    668.1 KB · Views: 91
  • 00276-1700121681.png
    00276-1700121681.png
    584.2 KB · Views: 96
  • 00289-1700121694.png
    00289-1700121694.png
    621.1 KB · Views: 87
  • 00198-3006756303.png
    00198-3006756303.png
    694.2 KB · Views: 90
  • 00185-3006756290.png
    00185-3006756290.png
    678 KB · Views: 102
  • 00155-3006756260.png
    00155-3006756260.png
    666.1 KB · Views: 100
  • 00152-3006756257.png
    00152-3006756257.png
    666.2 KB · Views: 95
  • 00301-1653137924.png
    00301-1653137924.png
    641.4 KB · Views: 95
  • 00275-1653137898.png
    00275-1653137898.png
    616.4 KB · Views: 96
  • 00249-1653137872.png
    00249-1653137872.png
    642.2 KB · Views: 89
  • 00219-1653137842.png
    00219-1653137842.png
    682.9 KB · Views: 88
Last edited:
Some impressions:

The "papercuts". Cool, but, for some reason, they freak me out a little--the bone color, the organic-like rococo gee-gaws. Anyway, onward.

Forrest tigers, deer, and robots. Unique, beautiful, complex and colorful. I begin to imagine a story. What kind of forrest is this? What kinds of creatures are they? Maybe they watch, observe, report on the astronauts.

The creature in the flowing white robes. A little like a dementor, but since it is white, perhaps the opposite. Great image and, especially I think, great composition.

The space ship. Did you design this, Nate? I, by now, have seen lots and lots of space ship interiors, but never one quite like this. I like it. The circles and ovals a little classic retro sci-fI.

The castle--reminds me of a kind of comic-booky image, but good.

The mall -- I had a scene, a flashback in a little thing where a guy has a breakdown in a futuristic mall. It might look something like this. Crowded, complex, claustrophobic. It just needs a soundtrack, Muzak, probably "Spanish Flea" :) .

The Astronauts in the foggy deserted town. Ok. These look great. And there is a story here, somewhere, An away mission into the Twilight Zone.

The pod-people. Nice design. Definitely gives the feeling of something bad going on here.

The elephants in the room. In the idiom they are meant to represent something too obvious to be ignored, but that is being ignored, for some reason, anyway. But here, they are seemingly being ignored because they are innocuous, maybe supposed to be there after all. I can see this image having some meme potential. Various elephants, various rooms, not being ignored because they represent something too difficult to talk about, but for some other reason--maybe they're service animals, bell-hops, I don't know, lol. Anyway.

The astronauts, helmets off, their faces. These are AI faces? Wow. Alive, expressive, with age and temperament appropriate creases, lines, and also with age and temperament appropriate absences of creases and lines. Realistic, and yet somehow, in some small way, still "Other." And the lighting in these is so bright and different from the others that it seems to tell some kind of, some part of, a story.

Anyway, also, nice glowey monkey, nice foggy space ship, nice all around. Bits of surrealism, magic realism, and hyper realism, and I wonder: could these three things somehow, logically, inhabit one story?
 
Last edited:
Some impressions:

The "papercuts". Cool, but, for some reason, they freak me out a little--the bone color, the organic-like rococo gee-gaws. Anyway, onward.

Forrest tigers, deer, and robots. Unique, beautiful, complex and colorful. I begin to imagine a story. What kind of forrest is this? What kinds of creatures are they? Maybe they watch, observe, report on the astronauts.

The creature in the flowing white robes. A little like a dementor, but since it is white, perhaps the opposite. Great image and, especially I think, great composition.

The space ship. Did you design this, Nate? I, by now, have seen lots and lots of space ship interiors, but never one quite like this. I like it. The circles and ovals a little classic retro sci-fI.

The castle--reminds me of a kind of comic-booky image, but good.

The mall -- I had a scene, a flashback in a little thing where a guy has a breakdown in a futuristic mall. It might look something like this. Crowded, complex, claustrophobic. It just needs a soundtrack, Muzak, probably "Spanish Flea" :) .

The Astronauts in the foggy deserted town. Ok. These look great. And there is a story here, somewhere, An away mission into the Twilight Zone.

The pod-people. Nice design. Definitely gives the feeling of something bad going on here.

The elephants in the room. In the idiom they are meant to represent something too obvious to be ignored, but that is being ignored, for some reason, anyway. But here, they are seemingly being ignored because they are innocuous, maybe supposed to be there after all. I can see this image having some meme potential. Various elephants, various rooms, not being ignored because they represent something too difficult to talk about, but for some other reason--maybe they're service animals, bell-hops, I don't know, lol. Anyway.

The astronauts, helmets off, their faces. These are AI faces? Wow. Alive, expressive, with age and temperament appropriate creases, lines, and also with age and temperament appropriate absences of creases and lines. Realistic, and yet somehow, in some small way, still "Other." And the lighting in these is so bright and different from the others that it seems to tell some kind of, some part of, a story.

Anyway, also, nice glowey monkey, nice foggy space ship, nice all around. Bits of surrealism, magic realism, and hyper realism, and I wonder: could these three things somehow, logically, inhabit one story?
Thanks for the feedback! I'm really happy to finally have the visual quality of the pipeline catching up to my imagination. I'd been trying to do transparent glass people for months, and finally got it working with some of the recent pipeline improvements. The human characters are almost ready for primetime, once the animation node is upgraded enough to accomplish more complex behavior.
 
After 9 hours of explaining the mechanics of general aesthetics to these brains, the pipeline now kind of understands stuff like the 60/30/10 concept, and how that can be achieved in any situation. So, checking off an important box on my to do list here "every frame a painting"


01054-283986480.png


01050-283986476.png


01043-283986469.png


01040-283986466.png


01034-283986460.png


01022-283986448.png


01021-283986447.png


01014-2035593133.png


01010-2035593129.png


01000-2035593119.png


00996-2035593115.png


00992-2035593111.png


00988-2035593107.png


00987-2035593106.png


00986-2035593105.png


00982-2035593101.png


00971-2035593090.png


00958-2035593077.png


00955-2035593074.png


00939-2035593058.png


00919-2035593038.png


00915-2035593034.png


00903-1083355971.png


00874-1083355942.png


00804-1037263119.png


00867-1083355935.png


00757-1037263072.png


00767-1037263082.png


00726-1037263041.png


01170-944060181.png


01151-672239936.png


01142-672239927.png


01111-672239896.png


01093-672239878.png


00473-272650348.png


00414-3080554476.png


00403-3182808140.png
 

Attachments

  • 00868-1083355936.png
    00868-1083355936.png
    553.2 KB · Views: 84
  • 00667-3598761224.png
    00667-3598761224.png
    558.1 KB · Views: 84
  • 01038-283986464.png
    01038-283986464.png
    653.4 KB · Views: 92
Back
Top