JoMME modifications

Get support for anything JK2 related, find tutorials for in-game and modding topics, and post any suggestions for the site here.
Tom Arrow
Posts: 92
Joined: 21 Sep 2016, 03:05

Re: JoMME modifications

Post by Tom Arrow »

Yeah, 1080 wasn't divisible by 100 perfectly, naturally. I used 108 blocks for this one.

Getting close!

Now the main problem really seems to be the performance so far. Hopefully it can be tackled. (Or Ill have to shut off AntiAliasing to make it workable)

Video version (15fps):
Attachments
vlcsnap-2016-10-07-08h10m54s738.jpg
vlcsnap-2016-10-07-08h10m54s738.jpg (313.16 KiB) Viewed 45193 times
User avatar
fau
Staff
Posts: 433
Joined: 16 Aug 2015, 01:01
Location: Warsaw / Poland
Contact:

Re: JoMME modifications

Post by fau »

If not, I suppose it will simply be slow, heh. But then we can forget about the pixel-based approach definitely.
You can forget about it using this exact method for sure, I explained it already. It's not speculation, it's estimation. Consider my sidenote idea if you want pixel-based approach, you can also interpolate between rows/rectangles instead of whole frames.

This screenshot looks strange, look at the jagged edge to the right. Are you certain time is incrementing in even periods? Do you use motion blur cvars to achieve fractional framerate? Maybe it's third person camera problem, can you try in first person (cg_fpls with saber)? I recall jomme used buggy third person camera code when rendering.

To improve performance examine SetViewportAndScissor function in tr_backend.cpp. Read OpenGL documentation on glSetViewport and glSetScissor. Limiting one of them to what is necessary there should hopefully speed up rendering, although I'm afraid with id tech 3 you're mostly cpu blocked at this point.

PS Where are these bespin textures from?
Tom Arrow
Posts: 92
Joined: 21 Sep 2016, 03:05

Re: JoMME modifications

Post by Tom Arrow »

Thanks, I'll check it out.

The jagged edges have a simple explanation really. I used 108 blocks. Each consists of 10 rows. Now, from the perspective when this frame started, the pillar on the right was leaning from top right into the lower middle of the picture. Hence in each 10-block, the direction of the line is left-bottom. But at the same time, the camera was moving the pillar out of the picture. Hence after every 10-block, there is a 'reset' to adjust for the new camera position. This should, I think, go away with full 1080 lines rolling shutter. Although you're right if you say that the camera may be a little wobbly. This would practically manifest as a wobbly-curvy edge rather than as a jagged edge though, unless there are jumps in the camera. Jumps in the camera occur often in 3rd person, for example when suddenly a wall is in the way of the camera. I don't think we can avoid this. Or maybe we could somehow smoothen that process, but frankly I am not interested in doing that, because I explicitly want a dirty look.

I didn't look at my CPU stats while recording, but I have a pretty high-performing CPU. Although it's multicore, so practically, only 1/12 of my CPU gets used I suppose.

The bespin textures are a little project of mine from photos inside of Munich. They are pretty high res and you can see finished cigarettes in the ridges, heh. I'll publish it when I have replaced all textures. Although I would need to learn to do some mapping to make it look good in all cases. For instance, bespin seems to be using the same texture for the ground on streets as for some walls, which, as you can imagine from the screenshot, leads to weird results.
Tom Arrow
Posts: 92
Joined: 21 Sep 2016, 03:05

Re: JoMME modifications

Post by Tom Arrow »

I suppose you could always set cg_thirdpersoncameradamp to a lower value if you want a smoother camera.
Tom Arrow
Posts: 92
Joined: 21 Sep 2016, 03:05

Re: JoMME modifications

Post by Tom Arrow »

Besides, jagged edges are very rolling shutter like:

Image
User avatar
fau
Staff
Posts: 433
Joined: 16 Aug 2015, 01:01
Location: Warsaw / Poland
Contact:

Re: JoMME modifications

Post by fau »

I don't mean jagged edges but how they are jagged. There are 2 small steps and a big one. Either time intervals between frames are uneven or it's a side-effect of broken third person camera code. Try in first person to make sure.
Tom Arrow
Posts: 92
Joined: 21 Sep 2016, 03:05

Re: JoMME modifications

Post by Tom Arrow »

I suspected you may mean that. I'll try if I have time, but I think it will become apparent when I use a full 1080 lines. Then there will either still be edges or the edges will be wavy, which would either point to some interpolation anomaly or simply to the way the movement was stored in the demo file. I am not sure if that is something that should be corrected. For one, I would actually like wavy edges. Second, if it was in the demo, it could not be fixed with better interpolation, but only with smoothing the complete data.
Tom Arrow
Posts: 92
Joined: 21 Sep 2016, 03:05

Re: JoMME modifications

Post by Tom Arrow »

Note that this was an offline demo also, so the time between packets may (depending on the code) be quite small and pick up even very subtle mouse movements that you could call 'measurement imprecision'. Not sure though.
User avatar
fau
Staff
Posts: 433
Joined: 16 Aug 2015, 01:01
Location: Warsaw / Poland
Contact:

Re: JoMME modifications

Post by fau »

Offline demo records as much data as online - server packets. The only difference is that if you were using non-dedicated server, server frames were not in regular 50 msec intervals. This accentuates third person camera warping and can be the cause of this – we won't know until you check.

I wrote a fix for damped third person camera that solves these issues some time ago but I'm not sending it to jomme until it's better tested.
Post Reply
Created by Matti from StylesFactory.pl and Warlords of Draenor (modified by jk2.info)
Powered by phpBB® Forum Software © phpBB Limited
 

 

cron