I feel like that number is almost completely meaningless because of how this works.
Imagine you were trying to break through a wall, And you needed 10,000 foot pounds of pressure to do it. If somebody told you that they were 90% of the way to achieving that, You would logically assume that they had 9000 foot pounds of pressure available.
What I'm doing is kind of hard to express, But it would be far more accurate To think of it as kind of a jagged exponential curve. I guess compound interest would be a simple way to say it. Like if you started with two foot pounds of pressure and then every 5% increased that by 1.55 X.
| **%** | **Value** |
|--------|--------------------|
| 0% | 2 |
Still images with cheap animation effects from off the shelf software
| 5% | 3.1 |
A more advanced version of the same thing With slightly animated characters
| 10% | 4.8 |
An even more advanced version of the same thing with stronger character animation and improved backgrounds
| 15% | 7.4 |
Beginning transition into 3D engines
| 20% | 11.5 |
Developing more advanced utilization of 3D engines for detail enhancement, Using cheap licensed music
| 25% | 17.8 |
It's now over a year in the development, And save point makes its first short film at the current level, Adding sound design to the mix for the first time
| 30% | 27.6 |
Begin expanding resources and biome availability in preparation for the eventual construction of larger worlds
| 35% | 42.8 |
Build on that with more advanced cinematography techniques designed to port forward into the upcoming ai layers
| 40% | 66.3 |
Push lighting and camera technique to provide an easier to recognize basis for future ai translation
| 45% | 102.8 |
Begin intended translation into hybrid 3D/AI space, Still working only on surface level, 1 week to complete A single short video of poor quality.
| 50% | 159.3 |
Moving forward with the second stage, Two weeks to complete a bad three minute video
| 55% | 246.9 |
Second stage beginning to stabilize and become coherent, A one minute video takes a week to produce, The save point pipeline is now well over two years in development
| 60% | 382.7 |
The process becomes coherent enough to create simple visually appealing scenes
| 65% | 593.2 |
The pipeline can now autonomously create higher quality themed worlds with a recognizable fictional setting beginning to emerge
| 70% | 919.5 |
The first small test product of a simple interactive game utilizing the stage two technology. Three weeks of work to produce 30 videos.
| 75% | 1,425.2 |
Beginning of simple storytelling, Visual fidelity improving, Minor speech integration, Quality is approaching minimum for broadcast
| 80% | 2,209.0 |
The first actual Save Point production cell, The first scene of "Labyrinth" Production time two weeks.
| 85% | 3,424.0 |
Artificially intelligent video automation achieved, Over 1000 music videos produced in two months.
| 90% | 5,307.2 |
A two minute video at four times the previous quality can be produced in one minute. There are now over 5000 videos, Construction begins on phase three, The final and most significant phase.
| 92% | 8,226.2 |
This is today at 92%. The project now contains tens of thousands of lines of original python source code. I could easily put ten feature length films here if I could spare the time to bother uploading them. But these are art films without a plot. Significant work is already complete on the process to reverse engineer and reengineer all media. Dynamically create new seasons of television shows that have gone off the air, Seasons six seven and eight of Breaking Bad, Season 78 and 9 of The Sopranos. Karasawa directs pulp fiction. Kubrick directs the A team with the original cast in 2027. Save point becomes the first infinite interactive film with tens of thousands of micro episodes driven by user choices. Those goals are now realistically within sight. On May 10th the Save Point pipeline watches and comprehends its first full length television episode, Dissecting every element of the 645 camera shots and correlating them to the episode script as it analyzes every dynamic creating the entertainment value, And begins its first glimmer of an understanding of the genetic code that might produce an infinite number of similar shows. Work continues on teaching the team of robot brains how to recognize characters roles tropes and every mechanical aspect that defines a genre or style it's exposed to. I begin teaching the system to correlate the length of individual camera shots to the dramatic effect they produce within the context of the plot structure and individual situations. The individual ai brains at work are now so powerful that they must be spread across network with dedicated environments, Offering analysis piecemeal through thousands of stages per job Because a one shot approach would require literally millions of dollars worth of vram.
The last 9% will be the most difficult and most important of all. Work is just one person putting in around 80 hours a week on average, The current goal is to have the system create a touring test capable episode of law and order before the end of the year. Accomplishing this simultaneously means that an unlimited Amount of new TV shows and movies can be produced with the same process.
As of now I still don't know whether it's possible in that timeline with no help and no investment, But what I'm already certain of is that all this is absolutely possible and right now it's only a matter of time.
| **100%** | **12,750.6** |
In the future, Save Point pipeline reaches version 1.0, I've broken through the wall, And on the other side is infinite possibility, And an opportunity to refine and polish a working technology that can supply a lifetime of entertainment to hundreds of millions of people around the world.
There are some really crazy possible developments after this point, Stereoscopic 3D is guaranteed, Having a conversation with a character in a tv show that influences how the plot turns out is feasible, Individual users tailoring shows is possible. Reconstruction of any movie or TV show so that you can stand inside it in virtual reality and watch it in a world happening around you is possible, Though at that point we were not even really talking about the television or film format anymore.
In any case I'd been too busy working to talk to anybody for a while, And since many have been following this project for years I thought some of you might appreciate an update.
Imagine you were trying to break through a wall, And you needed 10,000 foot pounds of pressure to do it. If somebody told you that they were 90% of the way to achieving that, You would logically assume that they had 9000 foot pounds of pressure available.
What I'm doing is kind of hard to express, But it would be far more accurate To think of it as kind of a jagged exponential curve. I guess compound interest would be a simple way to say it. Like if you started with two foot pounds of pressure and then every 5% increased that by 1.55 X.
| **%** | **Value** |
|--------|--------------------|
| 0% | 2 |
Still images with cheap animation effects from off the shelf software
| 5% | 3.1 |
A more advanced version of the same thing With slightly animated characters
| 10% | 4.8 |
An even more advanced version of the same thing with stronger character animation and improved backgrounds
| 15% | 7.4 |
Beginning transition into 3D engines
| 20% | 11.5 |
Developing more advanced utilization of 3D engines for detail enhancement, Using cheap licensed music
| 25% | 17.8 |
It's now over a year in the development, And save point makes its first short film at the current level, Adding sound design to the mix for the first time
| 30% | 27.6 |
Begin expanding resources and biome availability in preparation for the eventual construction of larger worlds
| 35% | 42.8 |
Build on that with more advanced cinematography techniques designed to port forward into the upcoming ai layers
| 40% | 66.3 |
Push lighting and camera technique to provide an easier to recognize basis for future ai translation
| 45% | 102.8 |
Begin intended translation into hybrid 3D/AI space, Still working only on surface level, 1 week to complete A single short video of poor quality.
| 50% | 159.3 |
Moving forward with the second stage, Two weeks to complete a bad three minute video
| 55% | 246.9 |
Second stage beginning to stabilize and become coherent, A one minute video takes a week to produce, The save point pipeline is now well over two years in development
| 60% | 382.7 |
The process becomes coherent enough to create simple visually appealing scenes
| 65% | 593.2 |
The pipeline can now autonomously create higher quality themed worlds with a recognizable fictional setting beginning to emerge
| 70% | 919.5 |
The first small test product of a simple interactive game utilizing the stage two technology. Three weeks of work to produce 30 videos.
| 75% | 1,425.2 |
Beginning of simple storytelling, Visual fidelity improving, Minor speech integration, Quality is approaching minimum for broadcast
| 80% | 2,209.0 |
The first actual Save Point production cell, The first scene of "Labyrinth" Production time two weeks.
| 85% | 3,424.0 |
Artificially intelligent video automation achieved, Over 1000 music videos produced in two months.
| 90% | 5,307.2 |
A two minute video at four times the previous quality can be produced in one minute. There are now over 5000 videos, Construction begins on phase three, The final and most significant phase.
| 92% | 8,226.2 |
This is today at 92%. The project now contains tens of thousands of lines of original python source code. I could easily put ten feature length films here if I could spare the time to bother uploading them. But these are art films without a plot. Significant work is already complete on the process to reverse engineer and reengineer all media. Dynamically create new seasons of television shows that have gone off the air, Seasons six seven and eight of Breaking Bad, Season 78 and 9 of The Sopranos. Karasawa directs pulp fiction. Kubrick directs the A team with the original cast in 2027. Save point becomes the first infinite interactive film with tens of thousands of micro episodes driven by user choices. Those goals are now realistically within sight. On May 10th the Save Point pipeline watches and comprehends its first full length television episode, Dissecting every element of the 645 camera shots and correlating them to the episode script as it analyzes every dynamic creating the entertainment value, And begins its first glimmer of an understanding of the genetic code that might produce an infinite number of similar shows. Work continues on teaching the team of robot brains how to recognize characters roles tropes and every mechanical aspect that defines a genre or style it's exposed to. I begin teaching the system to correlate the length of individual camera shots to the dramatic effect they produce within the context of the plot structure and individual situations. The individual ai brains at work are now so powerful that they must be spread across network with dedicated environments, Offering analysis piecemeal through thousands of stages per job Because a one shot approach would require literally millions of dollars worth of vram.
The last 9% will be the most difficult and most important of all. Work is just one person putting in around 80 hours a week on average, The current goal is to have the system create a touring test capable episode of law and order before the end of the year. Accomplishing this simultaneously means that an unlimited Amount of new TV shows and movies can be produced with the same process.
As of now I still don't know whether it's possible in that timeline with no help and no investment, But what I'm already certain of is that all this is absolutely possible and right now it's only a matter of time.
| **100%** | **12,750.6** |
In the future, Save Point pipeline reaches version 1.0, I've broken through the wall, And on the other side is infinite possibility, And an opportunity to refine and polish a working technology that can supply a lifetime of entertainment to hundreds of millions of people around the world.
There are some really crazy possible developments after this point, Stereoscopic 3D is guaranteed, Having a conversation with a character in a tv show that influences how the plot turns out is feasible, Individual users tailoring shows is possible. Reconstruction of any movie or TV show so that you can stand inside it in virtual reality and watch it in a world happening around you is possible, Though at that point we were not even really talking about the television or film format anymore.
In any case I'd been too busy working to talk to anybody for a while, And since many have been following this project for years I thought some of you might appreciate an update.
Last edited: