PDA

View Full Version : LW 9 - Caveat Developer



jin choung
08-16-2005, 08:08 PM
howdy guys!

woo hoo!!! i was holding out on a new upgrade until i got the things that i've been hollering about but finally, it seems that in 9, we'll finally get a revision of subdivision surfaces and edges and the edge weights!

so i took the plunge last week and am eagerly awaiting 8.5.

----------------------------------------------------------------------------

however, i wanted to keep a few things in mind for the developers that i think would be easy to overlook but would a tragic mistake if they did:

1. edge hard/ soft for gouraud shading. right now, whether polygon edges are shaded smooth or not is controlled inside of the SURFACES properties but it is not NECESSARILY a property of a surface/material.... there's no reason why you can't have user defined soft and hard edges within a single material.

since we need to create EDGE MAPS (ala vertex maps) for the edge weighting, we should incorporate such a system to desginate what edges should be smoothed over or not.

maya's is a great system to emulate (each edge can be tagged hard or soft manually). max's is awkward and unwieldy as **** with it's smoothing groups.... avoid that one like the plague!

2. RENDER GL!!! great, fantastic, yahoo! that you guys are utilizing the new generation of realtime graphics technology in todays graphics cards. but now, REMEMBER RENDER GL which was a great idea when it was incorporated iway back in the day but when it was not all that impressive.

but with the graphics cards on the market now, if you incorporate render gl now, you may be able to use it to render out FINAL FOOTAGE!!!

please please please don't forget!

mkiii
08-16-2005, 09:07 PM
I'll second that Jin.

Hard/Soft edges are a must. I am so tired of having to add extra geometry or adding yet another surface just to get a decent bevel etc. And god forbid, unwelding polys to create a crisp edge (which I almost always accidentally merge back together).

Even Max style smoothing groups would be better than nothing.

wacom
08-17-2005, 07:11 AM
maya's is a great system to emulate (each edge can be tagged hard or soft manually). max's is awkward and unwieldy as **** with it's smoothing groups.... avoid that one like the plague!



After using XSI I'd much rather have their way of doing it. It's always a right click away and you get to determine right away just how hard or soft the edge is (not just "on or off"). I'm sure NewTek can think of an even better way such as just using say ctrl-I and then holding down the mouse to slide between absolute hard to Soft.

jin choung
08-17-2005, 07:40 AM
hey wacom,

ummm, actually,

i'm not at all talking about SUBDIVISION EDGE WEIGHTING here. i'm just talking about gouraud smooth shading for polygon facets.

it's the thing in surfaces where you can click smooth and then define the degrees.... there is no (possible) degrees of sharpness here.

(oh yah, and there needs to be some graphical representation on the wireframe that tells us at a glance which edges are hard or soft... dotted lines vs. solid lines would work nicely - ala maya)

-------------------------------------------------------------------------
but i agree that fp values for varying degrees of sharpness is INFINITELY more valuable than the discrete hard/soft/semisoft options for maya. i mean WHY NOT give us infinite (more or less) degrees of sharpness?!

and also, considering edge weighting, the ability to determine weight and give usable feedback to the user will be challenging. the obvious way is by showing us a gradation on the 'edge map' but those things always look so frickin subtle to me... difficult to make out what's changing where!

jin

p.s. mkiii, i agree completely that unwelding is a PAIN IN THE *****!!! merging happens to be one of the things that i do at the end for model clean up and i always always always (!!!) end up rewelding stuff i didn't mean to! ACK!!!

it may be completely true that in fact, when an edge is considered 'hard' by a renderer, it is essentially unwelding and so lw gives us a modeling method that most closely resembles the actual render methodologies....

but for the love of pete, it's 2005, abstract it for us... please!

Meaty
08-17-2005, 08:12 AM
:agree:
Jin, as always, you hit the nail on the head! I know you have been a long time advocate of edge weights (back to the days of the old forum)... it certainly is nice to see them on their way.

dballesg
08-17-2005, 09:03 AM
Not bad ideas disscused here.

About RenderGL it was not created by Intergraph? But correct me if I am wrong, I think the company dissapeared donkeys years ago! :)

So I think that would be licensing problems, no?

Cheers.

Emmanuel
08-17-2005, 11:20 AM
If the edge weighting is as simple as dragging the mouse in the viewport to the left or right (AKA YO_weightslider right now for vertex weighting), I won't be dissapointed :)

Dodgy
08-17-2005, 11:20 AM
A little technique I use for keeping unwelded points unwelded. Make a Morph called 'Unwelded' and move the edge points away from each other. LW won't merge any points if they have different morph target positions, even if they're on top of one another in the base. Very useful in game dev :)

Emmanuel
08-17-2005, 11:23 AM
Yeah, if You consider it a FEATURE ;)
I had my problems with that feature, when finding out that I was modeling on the morph objects, instead of the base ....

archiea
08-17-2005, 12:54 PM
Not bad ideas disscused here.

About RenderGL it was not created by Intergraph? But correct me if I am wrong, I think the company dissapeared donkeys years ago! :)

So I think that would be licensing problems, no?

Cheers.


Its just a dot-com era statistic...

http://news.com.com/2100-1001-239083.html?legacy=cnet

jin choung
08-17-2005, 03:13 PM
hey dodgy,

thanks for the tip! i pray that we won't need such a workaround after nine but in case we do, i got something to work with.

as for rendergl, it ain't any proprietary dealy of intergraph's.... it's just stuffing the hardware render buffer to a file. really not a big deal. that's why i really want newtek to not forget it cuz not doing it would simply be a matter of neglect, not difficulty in implementing.

jin