Save Point V8 demo

Nate North

Nate North


CLICK THE COG BUTTON AND SELECT 1080p

This is a demo of the version 8.0 (or 80% complete) of the save point production pipeline. Upgrades can be seen sporadically throughout the reel and include:

Nuanced facial expressions
Hair blowing and wind
Color matching
New lip sync engine
Consistent characters
Automated foliage sway
Automated character blinking
Hybrid human/AI soundtrack music
climbing
etc.

Notes-

While only about 80 per cent complete at this point I feel like the pipeline is capable of producing saleable works.

Not all improvements are present in all scenes. A few shots here and there actually use the save .8 .1 and 8.2 pipelines.

The opening minute or two is the opening of the film death March in its current alpha state.

By the time the film reaches completion we are expected to have the technology ready to translate voice and lip sync, Allowing the film to be viewed in the top eight languages, with characters speaking the language on screen with matching mouth movements. That may or may not happen but it's looking fairly likely that it will.

Directing actors during line Delivery is finally working correctly, But has only been functional in the system for around a week so examples are limited within the reel.

The bottom line is that the system is essentially production ready. There is significant room for improvement and that will certainly happen as time passes, But as of last week It's possible to create an animated film on par with many existing professional release products.

The one issue slowing things down right now is maintaining even stylization across all scenes. It's not bad, and works most of the time but currently requires a lot of extra cutting specifically because of the delicate balance we're striking between Western Animation style and realism.
 
Last edited:
Upvote 5
Thanks everyone! I'm actually pretty happy with how this looks now. That having been said, I've made 2 significant upgrades in the last week. When I get a chance, I'll post a comparison clip here showing the change.

I think I've got all the technical stuff under control at this point, and my focus now is on developing believable acting performances. I feel like it's very very close now to where a few scenes actually hit the mark for the first time. In example, the "Nobody ever dies on their first mission" or "Sarge says we've only got 10 clicks to go" clips, feel to me like a par acting performances.

For me, this reel was the one where I felt like the characters began to feel alive, less like puppets and more like animated film characters.
 

There's forking versions now and have been for awhile, This short video was done as a test of one of the 8.1 forks. I'm in the middle of testing the 8.2 version right now, But I thought I would throw this up here just in case anyone was interested. This is just a small art piece "Clothworld III".
 
Thanks!

The difference you're seeing is in the mentality I'm approaching it with. Most see it as a toy but I see it as a tool. To be fair most of the implementations of AI available to the public are essentially toys, so I understand why people think of it that way.

Consider a musician that plugs a guitar into an amplifier and plays live vs wedding dj that plugs an MP3 player into an amplifier. In one case the person makes the music and the device amplifies it, In the other case the device makes the music. Same device, different method.

To me AI is only leverage and not at all a replacement for creativity. It lets me spend more of my time on the creative aspects, and less on expensive or repetitive tasks. In example, I wouldn't say there's a ton of artistic genius involved in deciding that ocean waves should lap up on the shore, literally any person in the world would automatically make that decision, so there's really no artistic talent involved in deciding that water should move on screen. And yet through manual methods it would consume potentially months of my time to animate a single scene by hand drawing each frame of a wave crashing onto a shoreline. That time can now be spent on legitimately creative tasks such as plot, pacing, and world building.
 

Here's what I think will be the final visual upgrade for awhile. I've just run the two main graphics cards in series allowing the pipeline to fill in about twice as much coherent small detail. When I get a minute I'll publish some before and after frames. I'm not sure if people can easily tell if this is a big upgrade, But if you look at before and after pictures closely, It's a big upgrade.

This 8.2 version essentially marks the completion of the original design spec from years ago. I have since moved the goal post quite a bit and there's a bunch of things that I want to accomplish with it in the future, Such as automating foley, Automating recreation of old films, etc, but at this point I have completed and delivered the original pipeline as promised.

So in short this video marks the completion of the original save point pipeline project.

I'm going to make this scifi movie Death March, Probably rebuild the Cat Labyrinth in record speed with the finished technology, Then begin building out the actual Save point game.
 
Last edited:
I'm going to use that hackneyed phrase about the journey almost being more important than the destination, but it seems to me that your journey with this project has been one of great leaps which paradoxically moved forward at a slow pace. It's amazing to me how far you've come with this pipeline and these techniques.
[Edit] Sorry if that sounds kind of flaky, lol.
 
Last edited:
Rusty Bolt said:
I'm going to use that hackneyed phrase about the journey almost being more important than the destination, but it seems to me that your journey with this project has been one of great leaps which paradoxically moved forward at a slow pace. It's amazing to me how far you've come with this pipeline and these techniques.
[Edit] Sorry if that sounds kind of flaky, lol.
Click to expand...
It felt like an eternity along the way, But it's been a genuinely interesting journey and more fun than most. The tech was moving so fast during the development. That there has always been some intriguing new possibility around the corner every step of the way so it's really kept me going especially the last couple of years. I think this would have been much harder to finish if I didn't have some new development to get excited about all over again every few months.

I feel pretty good about where things are at now mainly because the real journey is just beginning, since my destination these last three or four years was actually just the starting line of the race. Now it's finally time to do the thing I'm most excited about, make some new films and games!
 
It differs based on the situation, for example a weekend project would be charged hourly, but if somebody asked me to make them a 90 minute feature film there's a lot more involved and we would typically just set a project budget and try to complete the project within that budget.

In example if you ask me to make a 45 minute television pilot about an interstellar war, I need a lot of information to create a budget range with options including things like what percentage of the show is special effects shots such as space fleets or combat and what percentage is conversational.

Which aspects you are willing to do yourself will also play a significant role in pricing, In example if you have someone available to edit the finished film for you, that would no doubt save a significant amount of money.

My suggestion, if you would like to do some type of mercenary project with me, would be one of three options. You can make a very short film as a way to test the waters, build trust, or learn about the process. That could likely be accomplished for a few thousand dollars but you would almost certainly lose all of it as there is no market for films that short. The second option is to make a pilot, that could be 22 to 30 minutes 43 minutes to an hour, Dependent on your target market. Third, you could make a 90 minute feature film of some sort, and a 90 minute animated film is a fairly marketable product considering their rarity versus live action. This pipeline I'm working with has a new type of capability via which we can translate into any language including audio and the lip syncing of the characters making it appear native to a region. In any of the three above scenarios this technology opens up the opportunity to take the finished work into a greater than usual number of potential markets.

A great place to start would be for you to tell me a bit about what sort of project you'd like to execute, And we could begin a conversation about what that would involve..
 
Nate North said:
A great place to start would be for you to tell me a bit about what sort of project you'd like to execute, And we could begin a conversation about what that would involve..
Click to expand...

Nate,

I saw this video, which is part of a series compiling canonical bits of Star Trek with his sfx, and I thought that it was cool. For reasons that I will explain later, I may want to do a series of low-budget Youtube videos, which, statistically speaking may be better than expensive movies. Can you please look at the time 2.28 - 2.30, then at 5.46 - 7.07. How much would these cost?

 
These are standard blender scenes using available 3D models, I've done a few of these Star Trek fan fiction projects with the guys from "Star Trek continues". My pipeline is a completely different technology than what is used to create these scenes, Though it might work out really well for you. Basically I would usually charge around 7 or 8 grand to do "exactly" what you saw on the screen in Those sections. It would look better than that and you could be very detailed about the ship designs etc. That's using high end 3 D software, Which is what I used to use all the time before I built the save point pipeline. In my version you'd have considerably less control over the detail, But it would still look pretty good and cost maybe 10% as much.

To do exactly what you're seeing in those sections is a considerable amount of work. Depending on what visual sacrifices you were willing to make, it could be super cheap, like a few hundred bucks.

My standard rate for Viz FX is 85-110 per hour, dependent on the type of request. If paramount hired me to make those scenes for the big screen, it would be tens of thousands of dollars if not more. At a budget of 300 dollars, I could create a 1 minute section of animation that looked better than the sample below.

Just as an experiment, I spent an hour to see what type of ST fleet footage I could throw together in 1 hour of work.

So to give you a more useful frame of reference, here's that test, which would have cost about 100 dollars. Basically the more detailed you get the more it costs. In example, if you absolutely needed a specific logo on the side of one ship, I'd need to use train a new AI model (that's 3 grand) or use full 3d software and buy the models, then modify them, then film the scenes manually, so 100 dollars for a space battle, 10,000 dollars for a space battle with fine detail and a logo on the side of one ship. My point is, that it's an extremely non linear thing. I think the movie "battleship" spent about 70 million dollars getting about 10 minutes of footage exactly right.


This is the test I did, It took about 80 minutes, all inclusive. It shows midgrade animations of some ST ships and some space battles. Results would improve a good bit with more time. So if I spent 10 hours on the same 1 minute, it would likely be much stronger. If you made one very specific demand though, for example, there has to be this one moment where the starboard nacelle is hit by a beam and breaks off scattering debris into space, it's possible that that one shot could take me 10 hours to get exactly right.

Long story short, diminishing returns hits fast in CGI work, and twice as good or twice as specific can cost 10 or even 100x as much.

You should also be aware that scope and synergy play a big role in these jobs, so basically the smaller your scope the less effective your spend. Let's say you need 5 specific ships, and you do an hour show, and there are 8 scenes. That would only cost about twice as much as doing one scene, because the majority of the labor is in modeling and setup. On Star Trek TNG for example, we had the iridescent stars flying by, nebulas and clusters, etc. That probably cost millions to set up the first time, and then thousands each time they used it for the length of the series. So cost to do just the TNG opening credits would be in the millions, but cost to create 3 hours of footage like that across 7 years was probably only 5x that. (except it was more like 100 million, because hollywood vfx studios overcharge and grift like you wouldn't believe)

If you can accept the cheap version, something just a bit better than I showed above, it can be cheap.
 
Nate,

Thanks for the info, and I will need some time to absorb what you said. I can afford thousands, but not millions. With the democratization of the internet, the streaming services except Netflix have been losing money, and traditional shows like Late Night are going downhill, so Youtube and Instagram could be the way to go. The other way is to make tentpoles, and Deadpool and Wolverine just made over a billion dollars. So it seems that the middle - TV-like shows - is being hollowed out, with Youtube and the like gaining viewers on the low end and movies doing the same on the high end.

The situation doesn't seem to have stabilized, but at least for now, the low end can be lucrative.
 
You must log in or register to reply here.
Back
Top