Posted on 08/17/2020 6:59:02 PM PDT by UnwashedPeasant
both versions animations
Probably an animation but should be pretty easy to spit the flaws because 1902 computer graphics were not near as powerful as today’s graphics. /s
> You cannot “upscale film”. The pixels just aren’t there.
Soon computer modeling will be able to fill in plausible representations of missing details.
There is a lot of AI software now that can interpolate between pixels and intelligently tone and sharpen. Of course, if you start with actual film this is pretty easy as the resolution is almost continuous. But look at AI sharpening, noise reduction and resampling tools for digital photography and you’ll be amazed how credibly it can generate higher resolution images with almost no noticeable artifacts. This AI software is commercially available, reasonably priced, and in use by almost all serious photographers.
https://www.youtube.com/watch?v=AaDDd4Rixi4
Looks like "Steampunk" imagenings but is the real thing!
That technique is called bit stuffing.
The train is a unique design that was implemented in several German cities around 1900.
I cannot see anything in the entire video that looks fake - although the precision of motion is quite remarkable.
Anyway, whenever you see as much steel in one place as we see in that film, there is only one word that shouts out in your mind - Germany!
The city being filmed is Wuppertal, which was one of the first industrial cities in the world.
As to film quality...
Please remember that Germany and Switzerland led the world in camera and lens technology, probably right up to World War 2.
Very cool
Looks like an expensive project. I wonder if it ever paid for itself.
> fill in missing frames
And create plausible details missing in individual frames.
Some day I hope to see an old Humphrey Bogart film with such details generated.
When this happens, you will be able to zoom in to arbitrary levels of magnification and see generated details for example the wings of a bee sitting on a flower 25 feet away in the background of a scene, or stars in a night sky.
It will be used for evil as well.
Validating non-fake status of images has to be a growing area of practice.
Yes, plausible, but not necessarily accurate.
> Yes, plausible, but not necessarily accurate.
You can pick which way you want to view it, if at all.
Since the camera is moving, a 3D representation could also be generated.
About being accurate: known details from other photos of the subjects will be used in modeling less-detailed views some day, should the world last long enough.
For example you have a 16mm negative of a very old movie, but you also have sharp still photos of the actors. The details will be modeled from the stills.
Again, potential for good and evil.
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.