Regarding polygon count of character models

serializer

I mentioned this in a previous thread; basically, the polygon count of the character models is starting to look very low, to the point where there are obvious corners and edges to the characters, particularly noticable in areas like Max's ears and head.

When I talked about this before, someone (I think it was Yare) replied basically saying that they have been unable to increase the poly count because they need the games to still run on older computers.

Can I add some further thoughts to this debate now we are a few episodes in, and this issue is increasingly bugging me:

• The required specs for the game have increased anyway through much greater use of shaders
• Couldn't high-poly models just be used on detail settings 8 and 9?
• Especially in modern graphics cards, performance is much more related to the number of pixels, NOT the actual poly count. Obviously polys *slightly* affect performance, but increasing the poly count will have next to no impact on the number of pixels (therefore the total number of shader operations) required per frame
• As I said before, subdivision would be a very quick and cheap operation to smooth out the edges of the characters. This doesn't need to be done real-time, you can do all this in pre-process. I really think the performance impact would be negligible on modern computers, and would make a stunning visual difference to the game
• For the latest generation of graphics cards, you can instead use tesselation, which would look truly incredible with all the bump-mapping you now have. But high-poly models alone would run great on any half decent graphics card bought in the last 3 or 4 years!
• Did I mention that Sam & Max and other characters look really blocky, and this is the only visual aspect that marrs an otherwise amazing-looking engine!

I realise you're not going to make any major changes now half-way through a season, but please PLEASE bear this in mind for upcoming productions!