• While Stormtrack has discontinued its hosting of SpotterNetwork support on the forums, keep in mind that support for SpotterNetwork issues is available by emailing [email protected].

AI Editing Software

Warren Faidley

Supporter
Joined
May 7, 2006
Messages
2,739
Location
Mos Isley Space Port
The moderators might consider starting a new forum just for AI applications in chasing?

Has anyone found a decent AI software program that can handle 4k footage? I have yet to see any AI applications online that can handle 4k files, e.g., repairing exposure, blurring faces, removing defects/objects, etc. Many of these sites also rip you off by setting stupid limits on the amount of processing time you can use.

I do not believe this can be done online because of file sizes. The other issue is processing power. Most computers that are older than 7 years are not up to the task.
 
AI is becoming a substantial threat to water supplies as well, which is probably why they are not offering it to the public for free use, along with the massive data storage requirements on their end.



And these costs are being passed onto the consumer.
 
The moderators might consider starting a new forum just for AI applications in chasing?

Has anyone found a decent AI software program that can handle 4k footage? I have yet to see any AI applications online that can handle 4k files, e.g., repairing exposure, blurring faces, removing defects/objects, etc. Many of these sites also rip you off by setting stupid limits on the amount of processing time you can use.

I do not believe this can be done online because of file sizes. The other issue is processing power. Most computers that are older than 7 years are not up to the task.
AI (Algorithmic Illusion of Intelligence as I call it) is useful in some scenarios- mostly I hate it.

Topaz Video AI can do some of what you mention.

A full linear video editor with powerful frame processing is probably what you need though, and has a learning curve. I use Davinci Resolve with Fusion plugins for all things editing; the free version is quite powerful, and studio purchase is affordable and licensed non subscription. Some people like Adobe Premier. These programs have been doing complex edits with exposure and color correction, frame interpolation, object and noise removal, compositing, etc. for decades. They incorporate AI for things like improved denoise, but it is not automatically going to do all the work for you, nor would many want it to. You can do tons of things in a real video editor with a 3D effect add on, but it requires a lot of learning. All the online AI quick fix stuff may give you a free and instant result but will also keep your data, and you are paying some soulles company for their alleged IP that they built by stealing everything from the world to make that AI as "smart" as it seems to be.
 
AI (Algorithmic Illusion of Intelligence as I call it) is useful in some scenarios- mostly I hate it.

Topaz Video AI can do some of what you mention.

A full linear video editor with powerful frame processing is probably what you need though, and has a learning curve. I use Davinci Resolve with Fusion plugins for all things editing; the free version is quite powerful, and studio purchase is affordable and licensed non subscription. Some people like Adobe Premier. These programs have been doing complex edits with exposure and color correction, frame interpolation, object and noise removal, compositing, etc. for decades. They incorporate AI for things like improved denoise, but it is not automatically going to do all the work for you, nor would many want it to. You can do tons of things in a real video editor with a 3D effect add on, but it requires a lot of learning. All the online AI quick fix stuff may give you a free and instant result but will also keep your data, and you are paying some soulles company for their alleged IP that they built by stealing everything from the world to make that AI as "smart" as it seems to be.

Thanks for information. I do believe AI has a place in video editing as long as people don;t start adding major elements that were not part of the original frames. For example, you have a big bug crawling across the lens during the world's greatest tornado shot on a 20 minute, 4k clip. I'm not sure Adobe could handle that (a moving object) and fill in the background 100% accurately. It's coming, no doubt. The other problem is computing power / expense. My trusty, 2017 iMac 4.2 GHz Quad-Core Intel Core i7 is now a dinosaur in AI computing. To get where I need to be is a $4k+ expense w/o the monitor.

I would eventually like to covert old DV footage to 4k quality, again a job for AI.
 
I do use Topaz for editing photos. Their AI driven denoise, sharpen and upscaling can be very good especially for butterfly and wildlife images. For storms and dark sky, the results can also be really good or horrible. There is some trial and error. It's worth having. I really hate subscriptions and have stopped using Adobe except for my old version. I am learning Corel what has some AI features, but haven't had a chance to use them yet. I haven't used Topaz for video yet.
 
I tired using several AI programs (online versions) to remove horizontal shutter lines in older lightning footage, none of them came close. Nor could any of them successfully blur the faces of people moving quickly. We are still years away from AI 4k editing software where you can ask it to do specific tasks with 4k footage with 100% acceptable results.
 
I tired using several AI programs (online versions) to remove horizontal shutter lines in older lightning footage, none of them came close. Nor could any of them successfully blur the faces of people moving quickly. We are still years away from AI 4k editing software where you can ask it to do specific tasks with 4k footage with 100% acceptable results.
Are you describing interlacing issues with low resolution footage (looks like many bands with lines skipped)?

Or are you talking about rolling shutter where the bolt blows out or the sensor read is interrupted mid bolt by the next sensor read because the return stroke happens at such a fast scale the processor can'y catch it all? In that case the AI would have to regenerate a blown out portion of the image with made up guessed data. Nothing can do that yet that I know of.
 
Back
Top