veger / TLBE

TLBE - Time Lapse Base Edition
MIT License
9 stars 6 forks source link

Suggestion: Scaling Timelapse #56

Open DeadlyDad opened 10 months ago

DeadlyDad commented 10 months ago

Considering the 8-hour demo took 36GB worth of screenshots, a 180-hour run would require its own hard drive with the current technique. To get around this, I suggest adding an options to take screenshots A) every X structures placed (selectable minor structures--e.g. belts/rails/inserters/etc.--and major structures--e.g. refineries/assemblers/storage tanks/etc.) and/or B) every time the total number of structures (minor and/or major) increases by X%. The latter would have more screenshots taken at the very beginning when each structure is a larger percentage of the total and relatively few near the end. When compiled, this would naturally make the video speed up over time until it freezes with the rocket launch. (...and the camera REALLY should follow the rocket until it disappears!) This would make for a video that gets more and more exciting the longer it goes while minimizing the needed frame storage space.

Also, I would add an option to darken the area in the current frame that is identical to the previous one--barring belts/inserters/etc.--to highlight changes between the two.

Does that sound reasonable?

AestasLonewolf commented 10 months ago

Came here to suggest the exact same idea. Would be really nice to have the timelapse gradually slow down.

Maybe even just simply increase the interval ever-so-slightly with each screenshot taken.

veger commented 10 months ago

Hi thank you for your suggestions:

Considering the 8-hour demo took 36GB worth of screenshots, a 180-hour run would require its own hard drive with the current technique.

Each 30 minutes or so, I just convert the screenshots into a movie, which is only some 10s of MB large, and delete the screenshots.

The latter would have more screenshots taken at the very beginning when each structure is a larger percentage of the total and relatively few near the end.

I think it is the other way around, in the beginning you are usually low(er) on resources and hand-crafting (most) stuff. Later you get (almost) infinite resources, bots, etc. so the building goes much much faster. Also you'd require more buildings to reach certain crafting speeds/amounts.

This would make for a video that gets more and more exciting the longer it goes while minimizing the needed frame storage space.

I doubt it, in the beginning of timelapse videos you can see what is going on, building some mines, smelting arrays, first science(s), etc. Later, it becomes to much to keep track of, or existing things needs fixing.

Additional to this, the timelapse speed becomes very irregular, when building times moves fast(er) compared to when doing other stuff. This is clearly visible when looking at belts filling up (or getting drained). So I wonder if this will become 'annoying' (less satisfactory) for the viewer of the video (at least I always like to watch the transport lanes filling up or getting drained).

I think it makes more sense to play with the 'speed gain' setting during the game-play to speed up (or slow down) the video at certain points/milestones.

Also, with the newest mod version it is possible to pause the camera and manually take screenshots, this might work out for this situation as well? Although you'd need lots of screenshots to get a movie (instead of a slideshow), so it might be too much to manually properly.

Also, I would add an option to darken the area in the current frame that is identical to the previous one--barring belts/inserters/etc.--to highlight changes between the two.

The mod uses Factorio API to make screenshots, it is not rendering the game itself and store it as a screenshot. So post-processing the screenshots is not possible.

Instead, you'd need to find a video editor with some algorithm to detect (big) changes and darkening the rest. I have no idea if such tools exists though...

Does that sound reasonable?

The request is very reasonable on a technical level, but at the moment I have my doubts in the usability of it...