Could Dreamcast et al handle this/that game/effect? *DC tech retrospective *spawn

Actually that was done before , a handful of games overlapped blended quads . Best example I remember was skies of arcadia in the Opening scenes , they overlapped a quad with varying degrees transparency per each vertex to give it a soft light coming from an angle but it wasn't the only time used on dc. Overworks again used that same effect on the intro stages of Shinobi 2002 ps2, so it's a pretty simple but effective technique.

The guy who did soul calibur 2 test for dc mentioned that 60 fps dc is more delicate performance wise than it is at 30 fps. Also mentioned at 30 fps layered transparencies become negligible to impact performance. At 30 fps you can have scenes that range up 60k polygons per frame and 2 to 4 layers of transparent quad with almost no performance cost with them on or off. So imagine if they had been less 60 fps obsessed and more detail oriented at 30 fps since it has more performance legroom at that rate.

Fake bloom skies of arcadia
Screenshot-20240420-054425.png
Now that you mention the 30 fps, It is possible to lock the framerate to 30 fps in Doa2 keeping 1x speed, It would be a good test case to ramp the geometry of one stage + transparencies to the max and see how far we can push the console.
 
Now that you mention the 30 fps, It is possible to lock the framerate to 30 fps in Doa2 keeping 1x speed, It would be a good test case to ramp the geometry of one stage + transparencies to the max and see how far we can push the console.
That's interesting, so actual gameplay / physics animation don't slow down and sync just fine at 30 fps? Nice to see it's frame rate independent like that.

You might definitely be able to increase detail considerably then if you keep it consistent 30 fps. ( Might be able to use one of those Xbox 1 doa stages even) I wonder if one can change something in the 1st read to force 30 fps frame lock.
 
That's interesting, so actual gameplay / physics animation don't slow down and sync just fine at 30 fps? Nice to see it's frame rate independent like that.

You might definitely be able to increase detail considerably then if you keep it consistent 30 fps. ( Might be able to use one of those Xbox 1 doa stages even) I wonder if one can change something in the 1st read to force 30 fps frame lock.
No, It depends, when It is GPU bound It runs at 30 fps at 1x speed when It IS CPU bound It slows downs, or so It seems, but I found a memory address that locks the Game to 30fps.
 
Not really as you can see it's chugging in a lot of places and next to no collision on anything which frees the CPU up.

Pushing it farther than it can realistically go.
 
Last edited:
Not really as you can it's chugging in a lot of places and next to no collision on anything which frees the CPU up.

Pushing it farther than it can realistically go.

And collision i still there as well as phsyics( ik bones , gravity) nothing is freed up and these stages are higher in vertex counts than the default one. What youre looking at is mismatch between the visual model and the physics/collision model because he has swapped the visual one but not collision one yet( not sure if he has the tools to swap the collision models). not to mention he added material to stages( specular is heavy dc cpu) like in the desert stage the statues have specular if u look carefully. Theres MORE going on than on the defaut stage.

it shows you dont know how it actually works.
 
The stutter is from my capture card, I say it in the description.... That's why I usually capture off screen and I am reluctant to use my cheap capture card when I want to show how it runs on real hardware...

There is only one instance when it really does chugs for a split second. ( Soul Calibur 2 chugs on PS2 and I see no one doubting about its power...)

Also the game always has 2 stages loaded in ram, if tecmo improved the engine and made it load just one stage things could go beyond what they did. What at this point that is just fantasy, there is also a hard limit in the engine of the size of the fes for stages because between fights and custscenes it loads I. The background that's why the realtime intro switches between different stages and characters seemingly without loading times, except in the spiral stage.
If you mess with the ram in real-time you can switch between stages without loading but only from the current to the previous.
 
Last edited:
Unfortunately certain people still like to live in fantasy land and dream that DC was this super powerful machine that never reached it's full potential because PS2 killed it off.

BY 1998 standards it was super powerful though, that's simply how it is. The only thing more powerful were the Model 3 boards which were thousands of dollars each and even then the DC had some advantages. It was a lot faster than my Pentium 2 400 and Voodoo 2 - at least in terms of graphics (the Pentium 2 would have been faster at some simulation and scripting stuff no doubt).

It's just that technology was moving at a breakneck pace and two years was a long time then.

Arguably no machine ever truly reaches it's full potential, but having such a short life and limited development support meant there was little to no time to experiment with more advanced techniques that the system allowed, and assets were being made with primitive tools by developers who were still learning how to make best use of all the extra capability they'd gotten over PS1. The amazing Shenmue 1/2 for example was a Saturn remake made without automated version control, and bug management was done manually using an excel spreadsheet. Most of the assets for Shenmue 1 and 2 were actually made before the DC launched in Japan, and the only use of bump mapping was a seemingly experimental hand made application on a coin you looked at in a menu.

(Side note, but there's a great Shenmue Yu Suzuki talk with Mark Cerny translating, the game being as mindblowing as it was at the time is due to an insane vision, talent and raw determination. Mark Cerny being friends with Yu Suzuki is the icing on the cake).

If PS2 had seen only limited support and died off around the end of 2002 it wouldn't be the legendary system it turned out to be. PS1 software came on loads over the years, as did PS2 software. Somehow slap bang in the middle of these Dreamcast software wouldn't? This isn't a DC vs PS2 thing, it's a development thing.
 
BY 1998 standards it was super powerful though, that's simply how it is. The only thing more powerful were the Model 3 boards which were thousands of dollars each and even then the DC had some advantages. It was a lot faster than my Pentium 2 400 and Voodoo 2 - at least in terms of graphics (the Pentium 2 would have been faster at some simulation and scripting stuff no doubt).

It's just that technology was moving at a breakneck pace and two years was a long time then.

Arguably no machine ever truly reaches it's full potential, but having such a short life and limited development support meant there was little to no time to experiment with more advanced techniques that the system allowed, and assets were being made with primitive tools by developers who were still learning how to make best use of all the extra capability they'd gotten over PS1. The amazing Shenmue 1/2 for example was a Saturn remake made without automated version control, and bug management was done manually using an excel spreadsheet. Most of the assets for Shenmue 1 and 2 were actually made before the DC launched in Japan, and the only use of bump mapping was a seemingly experimental hand made application on a coin you looked at in a menu.

(Side note, but there's a great Shenmue Yu Suzuki talk with Mark Cerny translating, the game being as mindblowing as it was at the time is due to an insane vision, talent and raw determination. Mark Cerny being friends with Yu Suzuki is the icing on the cake).

If PS2 had seen only limited support and died off around the end of 2002 it wouldn't be the legendary system it turned out to be. PS1 software came on loads over the years, as did PS2 software. Somehow slap bang in the middle of these Dreamcast software wouldn't? This isn't a DC vs PS2 thing, it's a development thing.
Where can I find the talk?
Shenmue 1 I m convinced was very rushed and even has graphical implementations more akin to how the Sega Saturn worked. Shenmue 2 greatly improved but I am sure it was also rushed and if devs gave it more time it might have surprised us.

The games it got up until it's death werent exploring much its capabilities.
 
Where can I find the talk?
Shenmue 1 I m convinced was very rushed and even has graphical implementations more akin to how the Sega Saturn worked. Shenmue 2 greatly improved but I am sure it was also rushed and if devs gave it more time it might have surprised us.

The games it got up until it's death werent exploring much its capabilities.
 

Jesus! The shenmue Face demos have absolutely nothing on this!!! How did you get the lips and eyes to look so natural!!!??? @EsppiraK
That's because that's how the model was generated in metahuman before it was imported. After that its a matter of basic normal smoothing and textures. I m not sure if it was decimated before import to reduce polygons

Back then 3D artists had to manually model characters, without having the advanced sculpting tools we have today like Zbrush nor advanced software that generate models like Metahuman or Character Creator with proper anatomy and realism. They had to do that manually while having in mind a polygon budget that made sure it would run smoothly and be easy to animate. Plus they had to add the textures either through manual painting or manipulation of real photos (like Max Payne) that kind of matched the model's geometry.

Such model like this one was an impossible or nonsensical task for 3D game artists of the time and it would have required tremendous work to create and then animate like Irimajiri's head in that infamous tech demo.

The question is how many polygons is this model made of? It tanks the framerate
 
shenmue at the time of release made a huge impact graphically, too bad you're looking at it with today's eyes(but logical), at the time it was mindblowing.
I was also impressed when it was released. But retrospective observation can also reveal what we couldn't see back then. The DC's premature death, improvements made in Shenmue 2, comparisons with other games of that generation, knowing more about Shenmue's development and observing these graphical experiments are pretty telling about where the DC might have headed and what Shenmue could have been under different circumstances
 
That's because that's how the model was generated in metahuman before it was imported. After that its a matter of basic normal smoothing and textures. I m not sure if it was decimated before import to reduce polygons

Back then 3D artists had to manually model characters, without having the advanced sculpting tools we have today like Zbrush nor advanced software that generate models like Metahuman or Character Creator with proper anatomy and realism. They had to do that manually while having in mind a polygon budget that made sure it would run smoothly and be easy to animate. Plus they had to add the textures either through manual painting or manipulation of real photos (like Max Payne) that kind of matched the model's geometry.

Such model like this one was an impossible or nonsensical task for 3D game artists of the time and it would have required tremendous work to create and then animate like Irimajiri's head in that infamous tech demo.

The question is how many polygons is this model made of? It tanks the framerate
40K according to @EsppiraK on NeoGaf
 
Back
Top