Is 030 overated, and 040 underrated?

[eXeler0]

Originally Posted by d4rk3lf

Wow - that's amazing render mate - congrats!

I guess "scanline render" meant using the shadow maps, instead of raytrace, and that makes sense, because shadows maps needs to be stored in Ram like bitmaps, while raytrace is just math.
I assume, if you could make it run with shadow maps (if you had more ram, or made shadow maps les res), render would be waaaaay faster.

Anyway, awesome renders and work.

I've played a little bit with Lightwave in Winuae few months a go, but unfortunately, didn't had tiome to play more (same as you - becxause of regular fx 3D work), but I was impressed what various options Lightwave offered back then.
Hopefully, I?ll have more time in the future, because I really wanna put it to the limit, and see what kind of shot Amiga can pull of.

I asume Amiga could do shots in quality of Independence day, or Starship Troopers... just someone needs to make them

Thanx man,

Regarding render speed, the real performance killer was RayTraced soft shadows". Imagine did a brute force trick to simulate soft shadows.. it simply created a bunch of instances of the light with an offset from the original light to sort of simulate softer shadows. That feature alone murdered the performance ;-) 060 was the *only* option for me.

I used Lightwave a little and tbh it felt way more modern than Imagine, but I had already learned all the tricks in Imagine3d so I didn't wanna switch.
But quality wise, you could have gotten renders for space scenes in Lightwave where there are harsh shadows. Global Illumination algorithms weren't really a thing back then. (There was radiosity in some renderers but most didn't have it).
The part where things have improved a lot since the days of Babylon 5 is in digital post processing. Today, using modern post processing tools, you could have "saved/improved" the crappy renders in Babylon 5 in "post". ;-)

[eXeler0]

Originally Posted by VladR

Interesting project. I plugged this number to my excel sheet where I have benchmark data from Vampire (for all 3d pipeline stages) and it would render the scene in ~4 minutes using simple flatshading.

The 3D Mesh would use about 16 MB of RAM, so plenty of RAM is free.

Out of 15,000 frames (/60 = 4.1 minute), the 13,000 are used for 3D transform and quad set-up and 2,000 frames for scanline traversal.

Only 103 frames (1.5 second out of over 4 minutes) would be spent on pixel fill (actual drawing). Texturing would take slightly longer, shadows depending on approach also.


I reckon if I used indexing, it would drop to 25-50%, as edges are shared (between 2 polygons) and so are vertices (usually by 4 polygons) so no need to compute them again.


This particular scene - judging from wireframe - appears to have majority of polygons in chains and cobbles, so without them it could totally be real-time and playable [flatshaded] on Vampire as it should be under 30,000 tris easily.

Yea there are also alot of procedural textures in that scene. Imagine 3d included over 100 of procedural "textures" / materials that used very little memory compared to using bitmaps. Some of the complex ones rendered slow as hell compared to bitmaps, but again, the slow ones showed a lot of improvement in the jump from 030 to 060.

[d4rk3lf]

Originally Posted by eXeler0

Thanx man,

Regarding render speed, the real performance killer was RayTraced soft shadows". Imagine did a brute force trick to simulate soft shadows.. it simply created a bunch of instances of the light with an offset from the original light to sort of simulate softer shadows. That feature alone murdered the performance ;-) 060 was the *only* option for me.

Oh.. wow...
And I thought, I invented that (in 3DS Max).
I used that method in 3DS Max scanline render, with shadow maps, where I created small array of lights, with small distance between them, and they give sort of a very cheap area shadows.
I think I used it in this scanline render:
[ Show youtube player ]

But creating soft shadows out of bunch of (slow) raytrace shadow lights .... Imagine was very advanced back then since it offered that possibility, but it was kind of crazy, because that method would be very slow, even 10-15 later.

Originally Posted by eXeler0

I used Lightwave a little and tbh it felt way more modern than Imagine, but I had already learned all the tricks in Imagine3d so I didn't wanna switch.
But quality wise, you could have gotten renders for space scenes in Lightwave where there are harsh shadows.

Yeah, agreed.
What bothers me personally in Lightwave, is that the freaking modeler is totally separated as an separate app, and switching back and forth kind of bothers me.
But other then that (and some improvement in the animation keys, and UI), it really feel like a modern app... I am talking about LW 5 that runs on Amiga.

I've looked at some sample scenes, and yes, they had harsh shadows, but it can be easily fixed.
We have to remember, that back then, people didn't have same eyes like we have today . They did terrible mistake in lightning and texturing, without knowing it, and it looked realistic in their eyes, because it was a new thing.
They placed enormous amount of raytrace reflections EVERYWHERE, even it was a dirt ground, and they overused lens flare effects (well, some do it even today), and they used too much of procedural textures, instead of real photos (even in low res).. etc...
That's why would be very interesting to me, what would happen, if somebody try today to make best scene in these old app, with experience from past decades.

Originally Posted by eXeler0

Global Illumination algorithms weren't really a thing back then. (There was radiosity in some renderers but most didn't have it).

Global Illumination wasn't thing for me, even when they started to get very popular (2005-2010), and I personally avoided it for years.
There is some ways to fake it, and there was a really nice 3DS Max script E-light, that automatically creates sort of a dome, with lights (with shadow maps), pointing on the subject... it gave really GI look (with some small adjustments, and creating some additional lights).
Something like that I would try.
I saw some crazy realistic renders in scanline without GI, 10-15 years a go.

Well, your render is also proof how far you can get without GI

Originally Posted by eXeler0

The part where things have improved a lot since the days of Babylon 5 is in digital post processing. Today, using modern post processing tools, you could have "saved/improved" the crappy renders in Babylon 5 in "post". ;-)

Indeed.
But back in the day (Starship Troopers) they used a lots of miniatures combined with 3D models, and that really helped.
That's why I think that it's possible with Imagine or Lightwave (and some A4000) to create really impressive scenes not seen before, on Amiga.

[fxgogo]

I have been spending some time in Cinema 4D on the Amiga and I forgot how hard it was to light compared to today. We have been spoilt with GI, PBR and the other nice modern tools.

I always would get my younger colleagues and students to light with out the crutch of GI. The skills you learn are universal, even when you switch GI back on.

I remember when different editors were the rage. Imagine had it as well. But once I started using 3DS Max, I could not go back to the old way.

I really don't think I will be spending much time in 3D packages on my real Amigas. They are just too slow, even with my accelerator. It will all be emulation for me there.

[d4rk3lf]

Originally Posted by fxgogo

I always would get my younger colleagues and students to light with out the crutch of GI. The skills you learn are universal, even when you switch GI back on.

So true!
Also, even smallest practice in drawing and painting can drastically improve any 3D artist skills, in image composition, lightning, shading, modeling...

[Mathesar]

Originally Posted by grelbfarlk

Don't forget that the 040 FPU destroys the 030 68882 FPU in most situations.
Comparing an 030@50MHz vs an Apollo 25MHz 040 in AIBB in FP mode:

 name                        Savage    FMath    FMatrix    BeachBall Flops    TranTest FTrace    CplxTest

Blizzard 1230 III (030/50, 882/50, 3.0 in RAM)     197.32    27.48    6.08    36.97      66.64    92.63    119.57    7.32

Apollo 1240 (040/25, 3.0, OxyPatcher)         118.03    125.23    12.60    106.40      195.45 109.10    109.01    14.55

Blizzard PPC (040/25, 3.0 in RAM, 70ns)     118.33    125.26    11.61    80.87      195.52 100.09    109.34    14.57

Yes, floating point performance is much better on the 040 than on the 68882, even on 50 MHz.
I never did raytracing on my Blizzard1230/IV (with 68882@50) back in the day but I did do a lot of Spice simulations. Later I got a BlizzardPPC 040/25 and that was a bit faster.

However, the everyday experience and the 040 was slower than the 030...
I never liked it much for that reason. On paper it might have been faster, the experience was certainly not. The BLizzard 1230/IV on the other hand ran like a dream. I still think it is one of the best accelerators ever built for the Amiga 1200.

I think that nowadays you could design a decent 040 card. If you give it enough memory bandwidth and give it some cooling, why not?
Not that it would be useful nowadays but it could be fun building it
Maybe after I finish this 14MHz accelerator I might give it a try.

[eXeler0]

Originally Posted by Mathesar

Yes, floating point performance is much better on the 040 than on the 68882, even on 50 MHz.
I never did raytracing on my Blizzard1230/IV (with 68882@50) back in the day but I did do a lot of Spice simulations. Later I got a BlizzardPPC 040/25 and that was a bit faster.

However, the everyday experience and the 040 was slower than the 030...
I never liked it much for that reason. On paper it might have been faster, the experience was certainly not. The BLizzard 1230/IV on the other hand ran like a dream. I still think it is one of the best accelerators ever built for the Amiga 1200.

I think that nowadays you could design a decent 040 card. If you give it enough memory bandwidth and give it some cooling, why not?
Not that it would be useful nowadays but it could be fun building it
Maybe after I finish this 14MHz accelerator I might give it a try.

Yes, different CPUs for different use cases. I agree about the Blizzard 1230 it still is a great card (I have it my second 1200 but sadly it doesn't get much use these days). But if we ignore the FPU dependent apps a 50MHz 030 really is a great all around CPU when needed meaningful performance vs compatibility etc is taken into account.

That being said, I do think the full 040 is underrated when you look at how cheap they are compared to other CPUs of interest in the upper end of Amiga accelerators. If a simple but effective design could be done it would surely become a popular accelerator for many that want the whole package with CPU, MMU, and fast FPU.. Especially with say 128MB of fast RAM. (That would basically match a Mac Quadra 840 ;-)
Should you find the time and motivation to take on the challenge, people will sing song about you and I will make the effort of shipping a box of beer your way;-)

[Spriteer]

If Doom is some kind of measure I must say that CD32 with TF330 (68030@50Mhz) runs the game with acceptable speed. I dont have any Amigas with 68040 but I would put a 486 cooler or similar to it to keep it cool. Good processors both AFAIK.

[Seiya]

68030 boom is with Doom clone and to play good every 3D games also native for Amiga like Flight 3D simulator.

[kriz]

030 is great for mid 90s demos and some fps.. Had my blizzard 030 before getting a 060 in 1998 or around that... But in general a 060 (or vampire) is better all rounder the last 25 years lol