[Gambini]

Just did some tests with my map It lives.

I´m amazed. This is, with TROR, the most important thing added to Eduke32 in years. I can´t believe how well dynamic lights seem to blend with the shadetable shading. I tried to disable them with r_pr_lighting 0 but for some reason it didnt work.

Do you mind Plagman if I start to stalk you with reports and requests for the remainging issues distancing polymer to be the best renderer ever?

So far I´d love it to have a performance boost, voxel support, correct sprite clipping and increased draw distance. I know that´s not as easy as saying but if I fill your pm and email boxes ten times per day, you could have it done in a relatively short time.

[Micky C]

This is cool, some of those colours like in plagman's example were way off the original game using the old openGL shading system.

However from what I can see with the polymer lights, well they don't actually light anything. Sure if it's bright enough you can see the colours, but when it's pitch black the polymer lights don't make it brighter at all. Will this be fixed at some point?

Edit: Btw I'm impressed with how smooth it is. It seems where in classic the colours change suddenly in straight lines, here in polymer the colours fade in. Would still be interested to see what spherical shading would look like as opposed to radial.

[Mblackwell]

Gambini, on 07 May 2013 - 02:43 PM, said:

Just did some tests with my map It lives.

I´m amazed. This is, with TROR, the most important thing added to Eduke32 in years. I can´t believe how well dynamic lights seem to blend with the shadetable shading. I tried to disable them with r_pr_lighting 0 but for some reason it didnt work.

Do you mind Plagman if I start to stalk you with reports and requests for the remainging issues distancing polymer to be the best renderer ever?

So far I´d love it to have a performance boost, voxel support, correct sprite clipping and increased draw distance. I know that´s not as easy as saying but if I fill your pm and email boxes ten times per day, you could have it done in a relatively short time.

http://wiki.eduke32....er_Deficiencies

[DavoX]

I don't understand, what is going on? For real I can't see the difference in those pictures, maybe I'm just too sleepy and need to go to bed for once....

[TerminX]

You can't see the difference between the old and new pics? You can't be serious...

[fgsfds]

Something goes wrong with bloodtc.

Software Old polymer: New polymer:

All the textures with shade >30 turns pitch black: http://i.imgur.com/0WypGzq.png

Attached File(s)

•  bl_pal.zip (47.37K)
  Number of downloads: 453

[Plagman]

Does that use the BLOOD palette as-is? If so, any chance you can send it to me? I suspect the difference lies in the amount of shade offsets, and I was looking for a way to test that.

[Plagman]

Diaz, on 07 May 2013 - 11:38 AM, said:

By the way, I've been looking through the Polymer history pictures... are those reflective (mirror) surfaces with per-pixel opacity still possible, or have they ever been?

Yeah, the code is there, but to achieve the effect you need to put a specular map on a mirror, which I think isn't possible right now because Duke3D removes the mirror tile. If there was a way to define mirrors through CON you could.

[Plagman]

OK, I committed a couple of fixes in the last few revisions.

• Fixed interaction problem with lights where dark shades could not be lit
  
• Switched to spherical falloff; I don't think it's too shocking in comparison to classic after all, but if someone can make a compelling case maybe I'll add a toggle
  
• Generalized the algorithm to accept any amount of shades, which should fix issues with the BLOOD palette.

Spherical shading is actually more straightforward and saves some computation, so on top of looking more correct it's faster.

[fgsfds]

Awesome. Old problem with too much darkness in polymer seems solved. Thanks.

[Plagman]

Cool, glad it worked. I also just hooked it up to rotatesprite, so HUD stuff should now be properly affected.

[Plagman]

Micky C, on 07 May 2013 - 03:21 PM, said:

However from what I can see with the polymer lights, well they don't actually light anything. Sure if it's bright enough you can see the colours, but when it's pitch black the polymer lights don't make it brighter at all. Will this be fixed at some point?

That should be fixed now.

Micky C, on 07 May 2013 - 03:21 PM, said:

Edit: Btw I'm impressed with how smooth it is. It seems where in classic the colours change suddenly in straight lines, here in polymer the colours fade in. Would still be interested to see what spherical shading would look like as opposed to radial.

Yes, I specifically added code to interpolate between different shades over the distance rather than having strict thresholds like classic, which would be less expensive. It looks better that way, and I bet classic would have looked that way if it hadn't been prohibitively expensive to do that kind of computation in software.

[Paul B]

From what I've noticed it looks like Duke is carrying a flashlight to see far off in the distance this doesn’t really appeal to my eye. (For example: The further away walls are the brighter they become and as you approach the distant walls they darken? Tthis seems a bit backwards. (Could this be because I am using a maximum brightness on the sectors I am in?). I've been fooling around and fighting with shading for ages on my map I am working on and it seems like a never ending battle of adjusting shades to make it look properly. What would be the recommended sector shade for mapping? All I want is consistency. Some of the values i've been toggling are R_Shadescale 1.05 which seems to look the best and of course testing the sectors with R_PR_Lighting 0 or 1 just to make sure the lighting isn't interfering with the sector brightness.

Revisions are coming out like hot cakes and I can barely test them as fast as they come out! Exciting! =) I'll try the last release to see what happens with the draw distance. *** UPDATE *** Yeah i didn't see any corrections pertaining to what I am talking about.

I would like to point out there is a pretty big discrepancy between EDUKE Polymer shading and EDuke HRP shading. HRP shows the wall tiles not as dark and it seems to handle the shading better from a distance unlike just regular Polymer without HRP. Hopefully I am explaining myself properly.

Something else I just noticed. r_pr_lighting seems to turn on okay from the console in game if it is disabled. But once it's turned on, the disable "r_pr_lighting 0" doesn't take effect until you quit and reload the current map. Not sure if this is the intended behavior?

[Jblade]

This is very cool, although sadly this computer isn't up to using Polymer but it's pretty neat to know that it's much closer to classic now

I have a very minor request about the minitext command, would it be possible to add an orientation command to it? I can't configure it to widescreen settings to it looks off in all of the HUDs when in widescreen mode.

[Diaz]

Uhm. But isn't Polymer faster than software rendering if not using lights?

[Jblade]

Diaz, on 08 May 2013 - 12:37 AM, said:

Uhm. But isn't Polymer faster than software rendering if not using lights?

Uhm, no? I have an integrated graphics card on a laptop so there's no way I'd ever get better framerates than in classic mode.

[Micky C]

Diaz, on 08 May 2013 - 12:37 AM, said:

Uhm. But isn't Polymer faster than software rendering if not using lights?

When it's optimised, sure, but at the moment, noooooo wwwwaaaaaaaaayyyyyyyy.

Try firing up parkade.map or any large moderately detailed user maps, disable lights, and be amazed at the difference. Parkade is almost unplayable even without lights, and that's using the original 8-bit art, not the HRP.

[Diaz]

Must be that my graphics card is faster than my CPU then, but I could swear I got framerates in the thousands when running the original game with no lights nor HRP - didn't even get near that with software. My work computer is a different story though, here Polymer is much slower.

Polymost is faster than software on both computers, though.

[fgsfds]

Speaking of improving polymer. What about this http://forums.duke4....h__1#entry80706 ? It was there since polymer released and still not fixed.

[Micky C]

The change in visibility when firing a weapon is much more noticeable in polymer with the classic shade tables. It's like every single texture becomes fullbright, and it doesn't look very good. Fire the chaingun and it's like midday. That reminds me, are there still plans to get rid of this effect altogether in polymer, at least when lights are enabled? It's just so redundant and atmosphere-breaking, and weapon-spawned firing lights do the job so much better.

[DavoX]

I can, I thought there had to be a difference between the 2nd and 3rd.

[Mateos]

The BloodTC guys also posted a comparison here:
http://www.moddb.com...able-comparison

I think made this way it is easier to compare, as you can have all the 3 at once on screen

[Micky C]

It's definitely a lot closer, but how come that sprite on the wall on the left is yellow-ish when it should be maroon like in the bottom pic. And the floor texture has splashes of yellow when that colour isn't present in the original floor texture.

[Mblackwell]

Micky C, on 08 May 2013 - 02:11 AM, said:

The change in visibility when firing a weapon is much more noticeable in polymer with the classic shade tables. It's like every single texture becomes fullbright, and it doesn't look very good. Fire the chaingun and it's like midday. That reminds me, are there still plans to get rid of this effect altogether in polymer, at least when lights are enabled? It's just so redundant and atmosphere-breaking, and weapon-spawned firing lights do the job so much better.

Sure, reprogram the weapons and don't call "flash".

[fgsfds]

Micky C, on 08 May 2013 - 04:16 AM, said:

how come that sprite on the wall on the left is yellow-ish when it should be maroon like in the bottom pic. And the floor texture has splashes of yellow when that colour isn't present in the original floor texture.

Probably, some blood engine feature. This is how the same texture with shade 0 and 33 looks in eduke and blood. In eduke, it's still yellow when in blood all the yellow is gone.

Attached thumbnail(s)

• 

[Paul B]

Micky C, on 08 May 2013 - 02:11 AM, said:

The change in visibility when firing a weapon is much more noticeable in polymer with the classic shade tables. It's like every single texture becomes fullbright, and it doesn't look very good.

Even when not firing a weapon in Polymer, the textures further away look like they're being lit by a candle until you get close. The closer you get the darker the textures become. I'd say the lighting aspect still needs some work when using the Polymer renderer.

Well I tried classic mode on my laptop and the shading looks good. However in Classic mode I noticed that Duke doesn't see his reflection in the mirror. He is a ghost.

[Player Lin]

Wow, that's looking so cool for OpenGL renderer, more like the classic renderer colors...too bad it's polymer only...

[NightFright]

Folks, latest EDuke32 builds apparently broke the animated logo during startup (and possibly also the cutscenes, couldn't check that), it's just the first frame now. Using r3741 (with Polymer @ 32bpp).

[Plagman]

Yeah, I think I know what's going on, and how to fix it. I believe I know what the problem is with global visibility changes on shooting a gun as well, I probably just forgot to change the scale of the visibility when I changed it internally in the effect. I'll look at it tonight.

fgsfds: the issue you're talking about is the "doesn't save known good geometry" in the Polymer bugs wiki page. I'll get to it eventually, but it's definitely not first on my list.

[Gambini]

changelog says:

Quote

Revert to spherical shading; it's not like you could look up in classic anyway.

What was the problem with the first implemented vis method? I thought it worked great. You can look up in software and even if you dont do so, tall buildings being seen from certain distance are going to look different. I´m gonna do some visual tests but so far I don´t understand why changing it if it was working great.