Port Wolfenstein on Amiga 500 / 1000 OCS

[saimon69]

Originally Posted by Reido

Behind The Iron Gate?

Exactly that one!

I also remember the Justice playable demo including some textured polygons but speed on base 68k was slow and irrelevant for this topic

[d4rk3lf]

What do you guys consider vanilla A500? With 512kb Ram expansion, or without?
I personaly think, since these expansions were so largely used (I think 95% of the A500 had them), and many software requires it, I consider it as a part of A500. That being said, I wouldn't even considering porting Wolf3D with only 512kb.

@britelite
That was awesome man! Way beyond any of my expectation. I know there's no sprites yet and other stuff, but really cool to see how smooth it is on A500.
I really hope, you, or somebody take this, and go one more step further, in hope that maybe someday, it will be finished.
(even that Atari ST port is pretty good for me, if A500 can handle that speed (I'd call that a big success))

Cheers to all!

[kikems]

Originally Posted by d4rk3lf

What do you guys consider vanilla A500? With 512kb Ram expansion, or without?
I personaly think, since these expansions were so largely used (I think 95% of the A500 had them), and many software requires it, I consider it as a part of A500. That being said, I wouldn't even considering porting Wolf3D with only 512kb.

Cheers to all!

I think a standard base Amiga system as target for development today is A500 with 1MB ram ( 512 Kb chip + 512 Kb fast/slow ). If you can make the game for only 512Kb of ram you can say your game works in most of current A1000.

[britelite]

Originally Posted by d4rk3lf

What do you guys consider vanilla A500? With 512kb Ram expansion, or without?

512k+512k is pretty much the defacto standard nowadays

That was awesome man! Way beyond any of my expectation. I know there's no sprites yet and other stuff, but really cool to see how smooth it is on A500.

I actually had another look at the code and got it to run even better 😁 And there's still a few tricks that could be implemented.

I really hope, you, or somebody take this, and go one more step further, in hope that maybe someday, it will be finished.

If I ever continue working on this, I will probably not make it a wolf3d port, but rather a new game better designed around the 1MB A500.

[d4rk3lf]

Originally Posted by britelite

If I ever continue working on this, I will probably not make it a wolf3d port, but rather a new game better designed around the 1MB A500.

That's even more awesome, for me, personally.
Wolf is a good game, but I'd rather see something new.
btw...here's a similar topic from a really nice and knowledgeable guy ReadOnlyCat:
http://eab.abime.net/showthread.php?t=79538
EDIT: oh... I see you participated in that thread already... well... maybe someone will find that thread interesting

Anyway, it's really cool to see how far A500 can go with doom like clones.
One little off topic question:
Would A500 better (faster) handle gradients then textures on 3D models? I have in mind something similar like "Behind the iron gate", but not exactly like that. More like a fake of ambient occlusion.

[britelite]

Originally Posted by d4rk3lf

Would A500 better (faster) handle gradients then textures on 3D models? I have in mind something similar like "Behind the iron gate", but not exactly like that. More like a fake of ambient occlusion.

Untextured walls with depth shading would indeed be a lot faster on A500, as they wouldn't need to be drawn on a per-pixel basis.

[Sir_Lucas]

Originally Posted by britelite

As I don't really enjoy being called a liar, I thought I'd share a little sneak peak of what could have been.



Sprites were never impemented and the wall rendering could still be optimized further. The wall rendering is done with vertical stripes of unrolled code for each zoomlevel and textures are 64x64 in size. Rendering is done in 2x2 resolution, but the blitter c2p routine used allows for two colors within each 2x2, giving free dithering. The ceiling/floor is currently being cleared with the blitter, and possible game logic would probably be performed during the blitter operation.

Last but not least, an non-interactive executable, runnable on a standard A500 512k+512k configuration:

http://www.dekadence64.org/wolf3d.lha

This project has now been shelved for my part, but if anyone else is working on a wolf3d-port, then I'll be glad to help.

If anyone is interested in discussing the technical part, I can start a new thread for that.

This is way too cool to leave it as is. Great job!!

[d4rk3lf]

Little of topic (not connected with the wolf3d port)

Originally Posted by britelite

Untextured walls with depth shading would indeed be a lot faster on A500, as they wouldn't need to be drawn on a per-pixel basis.

Thanks.
Some idea is on my mind for some time, and I did a quick scene in 3DS Max to explain my thoughts. I have no idea about coding, of course.
You will see a scene where I kind'a faked global ilumination, but all that is using is gradients on polygon, and polygons with just a pure color.
I attached the picture of polygons used (all is done with them).
Is Amiga capable of something like this? Is this right approach of thinking?
Of course, this scene is very very basic.

Thanks in advance.

Cheers to all.

[Megol]

Originally Posted by d4rk3lf

Little of topic (not connected with the wolf3d port)



Thanks.
Some idea is on my mind for some time, and I did a quick scene in 3DS Max to explain my thoughts. I have no idea about coding, of course.
You will see a scene where I kind'a faked global ilumination, but all that is using is gradients on polygon, and polygons with just a pure color.
I attached the picture of polygons used (all is done with them).
Is Amiga capable of something like this? Is this right approach of thinking?
Of course, this scene is very very basic.

Thanks in advance.

Cheers to all.

It should be possible to store pre-scaled slices and use the blitter to draw them.

But the floors and ceiling would most likely need to be one color.

[Galahad/FLT]

Originally Posted by Megol

It should be possible to store pre-scaled slices and use the blitter to draw them.

But the floors and ceiling would most likely need to be one color.

In that case, be better off "drawing" the floor and ceiling with a copperlist gradient, a dynamic one like as was used in Corporation and Robocop 3

[hooverphonique]

Originally Posted by d4rk3lf

Little of topic (not connected with the wolf3d port)



Thanks.
Some idea is on my mind for some time, and I did a quick scene in 3DS Max to explain my thoughts. I have no idea about coding, of course.
You will see a scene where I kind'a faked global ilumination, but all that is using is gradients on polygon, and polygons with just a pure color.
I attached the picture of polygons used (all is done with them).
Is Amiga capable of something like this? Is this right approach of thinking?
Of course, this scene is very very basic.

Thanks in advance.

Cheers to all.

Each vertical line of a wall would need to be a single color (i.e. a wall could have varying color left-to-right, but not top-to-bottom). What you're suggesting there is pretty much the same as texture mapping as each pixel needs to explicitly placed by the cpu (unless the shading somehow can be done using the copper as mentioned, but that could get very complicated).

[LaBodilsen]

Originally Posted by britelite

As I don't really enjoy being called a liar, I thought I'd share a little sneak peak of what could have been.

(picture removed)

Sprites were never impemented and the wall rendering could still be optimized further. The wall rendering is done with vertical stripes of unrolled code for each zoomlevel and textures are 64x64 in size. Rendering is done in 2x2 resolution, but the blitter c2p routine used allows for two colors within each 2x2, giving free dithering. The ceiling/floor is currently being cleared with the blitter, and possible game logic would probably be performed during the blitter operation.
.

This is running so much better than i ever would have imagined it doing with a C2P approach, and to me clearly shows that it is possible to get a Wolfenstein like game running on A500 with acceptable framerate.

I hope you don't mind that i have decrunched and disassembled the code for educational purposes (and have it running great in Asm-one by now).

Looking at the code, it's not even heavyly optimized, apart from the unrolled scalar code, and as you say yourself, it can be faster (evident by the fact that you have made it faster after this demo)

If anyone is interested in discussing the technical part, I can start a new thread for that.

It does open up a lot of questions for me, but i will wait with these until i have studied the code some more. But i do think a thread on the technical aspects would be great.

[britelite]

Originally Posted by LaBodilsen

I hope you don't mind that i have decrunched and disassembled the code for educational purposes (and have it running great in Asm-one by now).

I wouldn't had released the binary if I'd mind someone having a look at it

Looking at the code, it's not even heavyly optimized, apart from the unrolled scalar code, and as you say yourself, it can be faster (evident by the fact that you have made it faster after this demo)

Indeed, this was mainly a testbed for the wall rendering, so other parts aren't really optimized.

It does open up a lot of questions for me, but i will wait with these until i have studied the code some more. But i do think a thread on the technical aspects would be great.

Alright, I'll probably open up a thread later today when I get home.

[d4rk3lf]

@LaBodilsen and britelite
That's just fantastic. Looking forward to see any work you done. Good luck!

@LaBodilsen
Would you attempt to make a wolf3D port, or some your own game like britelite, or just for styding and learning purpose, that may (later) leads to one of the above options?
Just curious thou, no need to answer if you don't want to.

Originally Posted by hooverphonique

Each vertical line of a wall would need to be a single color (i.e. a wall could have varying color left-to-right, but not top-to-bottom). What you're suggesting there is pretty much the same as texture mapping as each pixel needs to explicitly placed by the cpu (unless the shading somehow can be done using the copper as mentioned, but that could get very complicated).

Thanks.
Well, I was hoping, if we can use some procedurals (gradients or whatever) instead of textures, to make something run really fast on A500. Unfortunately, my knowledge is very limited to offer any useful idea.
Anyway, I was always wondering, why programmers (almost) always insists on textures, even when they try 3D shooter on 8-bit machines? Why not just pure color, but fast and nice gameplay? Something like behind the Iron gate. There's many attempts with textured games, and very few untextured.
Cheers!

@Galahad/FLT and hooverphonique

Thanks and cheers!

[Megol]

Originally Posted by hooverphonique

Each vertical line of a wall would need to be a single color (i.e. a wall could have varying color left-to-right, but not top-to-bottom). What you're suggesting there is pretty much the same as texture mapping as each pixel needs to explicitly placed by the cpu (unless the shading somehow can be done using the copper as mentioned, but that could get very complicated).

The difference is that with a limited palette and limited types of gradients it would be possible to store pre-scaled slices reducing the inner loop to a simple copy.

That could make a huge difference.

[britelite]

As promised, I started a new thread for more technical talk.

[LaBodilsen]

Originally Posted by d4rk3lf

@LaBodilsen
Would you attempt to make a wolf3D port, or some your own game like britelite, or just for styding and learning purpose, that may (later) leads to one of the above options?
Just curious thou, no need to answer if you don't want to.

I dont intend to make a Wolf3D port per se, so for the time beeing it's only for learning purpose, as i've always wanted to try and write a 2.5D raycaster myself.

[utri007]

Originally Posted by britelite

If I ever continue working on this, I will probably not make it a wolf3d port, but rather a new game better designed around the 1MB A500.

I have seen lots of nice demos, but very few of them has actually materialized to games. It is has much less work to port game than create new.

Creating all the sprites and levels is big job. Bigger than many could though.

Another thing is, can you keep fps if there is a movin object in screen, like enemy?

[LaBodilsen]

Had a little fun today, and spent sometime to understand the texture format of britelites demo.

So.. This is how i remember Wolf3D looking. (sorry for bad texture conversion)


Also added different greys for floor and ceiling.

[Galahad/FLT]

Originally Posted by LaBodilsen

Had a little fun today, and spent sometime to understand the texture format of britelites demo.

So.. This is how i remember Wolf3D looking. (sorry for bad texture conversion)


Also added different greys for floor and ceiling.

That looks surprisngly effective, and a virtually free implmentation as well. Gives it a hint more depth.