Anyone currently developing on a Katana?

Just wanted to say, awesome work there.

-hl718

 

i would like to make my own simulator would be good but i may start on something small like a space invader type game then go for the bigger type games

 

is there any freeware 3d software

 

Blender

 

Excellent suggestion, they even feature some tutorials on their website:

http://www.blender.org/education-help/tutorials/

 

Tokyo Bus Guide yeh but it would be ok if i could get it in english thanks

 

Wow Vitamen, that's the only thing I can say. Impressive.

Awesome work, when will you show us some more ( video maybe )?
Do you code this on a katana?

7 times DOA2? That's really a good performance.

I can't wait to see a powerful homebrew 3d engine for the Dreamcast.

 

yeah about 7 times the amount of geometry featured in DOA2, but i do not do the type of animation they did, i just use DOA2 because it is the most geometry heavy dc game i have seen. I wish i had in my hands some type of detailed animation so i could develop an optimized format for it...

anyway, i do not know when i am going to work on dc again, right now i have other serious matters to take care of, like find a job and stuff like that. I think sometime next year i am going to have a 10 day burnout session to develop the new pipeline i designed and which will be even faster (but mind you, the dc's small vram won't be able to hold the extra geometry ).

 

Great, well I will look forward to that, good luck in the rest of your matters.

But that's what the dc really needs,a powerful homebrew 3d engine.

 

wow good job.Iam curious though, I remember on makaron developer blog some one asked deunan about the triangle counts and deunan replied that trigger heart does over 33000 verts and over 44000 triangles per frame at 60fps.What i wanted to ask you were you that anonymous poster that claimed with careful programming the dc can do 4.1 million triangles(60 something thousand per frame) at 60 fps with 1 light?

I was just asking because iam curious about that and it seemed like too much of coincidence.Also wouldnt trigger heart top dead or alive 2 on pure polygon pushing by deunans stats?What about the dc doing 3 million tri at 30fps? wouldnt that double the detail and put less strain on it?I would love for you to release a polygon pushing tech demo =).

 

no, the anonymous poster was not me...from the quote you posted, i guess his was a simple test demo that worked well with the optimally stripped tori (optimally stripped meaning with each extra vertex you have one new polygon using the previous two). Spheres, tori and simple geometry like that can be reduced to strips very efficiently. Unless you have a tool that can stripify real game geometry effciently, the 4.1 mpps number is a bit misleading.

On my demo i worked with real meshes like characters, cars etc...and my exporter cuts the geometry size in half with zero detail loss when stripping them. So i have 80000 verts / 3 = 26666 real triangles per frame with one light and one texture. If i enable stripping you can double this number so it is 53333 "effective" triangles per frame. At 60 fps with stripping enabled you get 53333 * 60 = 3.2 million triangles per second. Along with a 640x480 24bpp framebuffer, you are left with 1.2 megabytes for textures, which is not much. So the whole concept is to make it superfast to free cpu time, since if you push too many vertices you won't have enough space for textures...

There are indeed a number of tricks to free up vram space, most notably strip buffers but from what i have heard and understand it is a solution that would be very unstable in real games, producing everything from rendering errors to slowdowns, so i haven't really played around with it.

I will try to release the demo in the near future...add some better controls for the camera and maybe implement environment mapping as well....

 

hmm How does performance of windows ce compare to sega's own developing libraries? Always been curious about that.

 

well, i do not really think that using a certain sdk automatically makes your game slower or faster...provided that the wince sdk lets you get really close to the hardware, then i do not see any reason for not letting your engine befast...

i have no experience with the ce sdk, but i think that the katana sdk was better because it was pretty minimal and let teams integrate it better into their engines...i believe that the wince sdk is perceived as slow because dev companies tried to port their win32 code with minimal effort...

i have also started slowly cleaning up my code, i will try to make an interesting demo within this year, let y'all know about it when everything is ready

 

So in the end all the terrible wince games was based on laziness and not the actual quality of the sdk?I remember a few devs complained about really early versions of windows ce on the dc being really basic and crude to the point of not working right to not working at all.The only two devs that come to mind are the creators of draconus,treyarch (they claimed it wouldnt work for them at all) and the devs of slave zero(they couldnt figure how to stream sounds/bgm while playing sound effect).

Most games have framrate problems,even segas own smilebit couldnt lock the framerate of sega rally 2,though they seem to prefer developing using direct x, odd enough.(sega really 2,hundred swords,typing of the dead pc version, and their future game gunvalkyrie using windows ce.)

Glad to hear a demo is on its way and POSSIBLY this year.Looking forward to it.

 

i will release the one i have ready, i just need some time to cleanup everything, add a few controls and choose an interesting mesh...

 

Looking forward to your release. It's great to see people still working on stuff.

 

I wish i could code on a Dreamcast

 

i have some questions:

1) can the katana sdk be used without having a devkit?

2) how do you solve cache misses? i've read some kalisto docs that speaks about reordering loops and improving data alignment, but that's all that i've seen. There's another trick to get more performance? (like using sh4 asm cache manage instructions)

 

Yes you can. You can test your program using an emulator (nulldc or else Makaron) or using dcload-serial/dcload-ip. There're just few things to adjust (editing the linker script...). If you need help for that, maybe I can prepare a "sdk toolbox" for you.

I don't have read this thread well. You have trouble with cache? In the sdk, you have some methods concerning cache, you should have a look to them.

 

oh, ok. I haven't read the sdk docs, in fact i just use kallistios. Will try to take a look on that. Thanks!