Quake3World

spookmineer wrote:If you want to make quality movies, stay away from Fraps... You need uncompressed quality untill the final compression.

spookmineer wrote:No it doesn't... I don't know if Fraps uses jpeg which is already compressed, but I did a test to make sure, recording with Fraps for 1 minute @ 800x600 is 0.99 GB. The same minute (same demo) using cl_avidemo 25 results in 2.11 GB. The reason being cl_avidemo makes screenshots in TGA format which is truly uncompressed, while Fraps may be using "uncompressed" jpeg which to begin with is always compressed.

In any case, there are plenty of movies recorded with Fraps that just don't look as good as movies using cl_avidemo (or dumpframes for UT2k4). Movie makers who want an end product that looks like the game, stay away from Fraps.

For more info, http://www.own-age.com

Fjoggs wrote:www.own-age.com contrains everything you need.

Foo wrote:Was Fraps recording at the same FPS?

BTW I'm not arguing for using Fraps, just curious.

+JuggerNaut+ wrote:

Fjoggs wrote:www.own-age.com contrains everything you need.

old

just did a huge number of test run avidemos with different settings and counted the number of outputted tgas in each case. What i found was interesting.

i did a cl_avidemo of a single demo with avidemo values of 1,2,10,20,40,125,250 each with timescale 1, timescale 0.1, timedemo 1, and each with high quality and low quality settings. I even did tests with demos that had been recorded with sv_fps 15, sv_fps 125, sv_fps 20. I even did demos that had been recorded with g_sync.. 1 and left at 1, and demos that had been recorded with g_sync 1, and then turned to 0 after starting the record.

here's what i found:

the number of outputted frames per second is not limited at all. I did a test with cl_avidemo 250 and it outputted 1782 tgas, whereas with cl_avidemo 20 it outputted 142 tgas.

The timescale that is set while making the avidemo is inversely proportional to the number of tgas outputted (but since the actual running time of the resulting avi demo is also inversely proportional to the timescale, the outputted fps of the avi is unchanged.)

ie:

say you have a demo that is 1 second long.

with cl_avidemo 10, and timescale 1, you will output 10 tgas. Now if you string these tgas into an avi, the running time of the avi is 1 second long. Thus it is 10 frames per second.

Now say you decrease timescale to 0.1; you will now output 100 tgas. The running time of the avi is now 10 seconds.

100 tgas in 10 seconds = 10 frames per second.

i even tested with timescale 0.01 and sure enough it outputted 100 times as many frames as timescale 1. and therefore slowed the avi down by a factor of 100.

*note i never actually stringed the tgas together but i'm almost positive that the running times would be scaled down. (i have stringed an avi that was made i timescale 0.1 and it was very slow motion)

Now with timedemo 1, it doesn't matter WHAT timescale you use, or what value of cl_avidemo you use, the outputted number of tgas will be the same. And i'm almost positive that it's 20 tgas per second of running time = 20fps (which is probably why Dr Watson experienced 20 fps even with a cl_avidemo value of 1)

in my case i was getting 141 tgas with timedemo 1, and the demo was about 7 seconds long : 141/7 is about 20 fps.


EDIT: forgot to add that the quality of visual config you use while making the avi demo does not impact the output number of tgas - it does impact the time it takes to create the avidemo however.

EDIT: it's possible that sv_fps limits the output fps when using timedemo 1. i didn't test this thoroughly but i don't think that this is the case.