Tailor Archive

Since the devs announce their plans for an “Urban Planner” elite-elite profession based on Master Architect, and their stated plans for more elite-elite professions, I’ve been thinking about an elite-elite profession based on Master Tailor.

First off, the ‘easy’ route for the devs here would be just to make another tailor profession with more schematics. See my response to the “Clothing Wishlist” post (http://forums.station.sony.com/swg/board/message?board.id=tailor&message.id=5564) on how they could do this.

But frankly, that lacks all imagination and to be honest I don’t think that really gives us what we ‘want’ (well, next to more schematics of course ). So I’ve been thinking about what it really is that we want (or at least what I think we want). And the answer? We want to design our own clothes. We don’t just want to put some resources together and voila, a shirt. We want to specify the sleeves, the collar, the fabric, the buttons, etc.

So I’ve come up with an idea for an elite-elite profession called the “Fashion Designer”. Right now I was thinking about it consisting of the following four branches:

Go and check out THERE, they'replayer designed clothing, though in avery different way,and I know that Raph is aware of the game because he's complemented how well their emotes are done.

I would love something like this and I think it could be done.

The downside is that SWG already has a database problem and I'm going to guess this would make it far far worse.

Great suggestion .... I like it because it truly adds something new to the game rather than just more cookie cutter clothes templates (not that more choices are bad )

But as the other poster mentioned, I hate to think of the database problems this will cause ... *shudder*

I'm not sure it would be that big a database issue. Instead of having 20 slightly different long-sleeved shirts and 5 different short-sleeved shirts, you'd have one base shirt template, two sleeve templates, three or four neckline templates, a few button patterns.... if you wrangle it right, you might end up making the database more efficient, not less.

I want this functionality, but I don't want any "elite elite" professions. I'm only halfway up the tailoring tree, but I'm already out of skill points and making painful choices as to what I should keep and what should be trimmed. With only one character per server, "elite elite" professions essentially turn us into one-trick ponies, even if it's a really good trick. That's bad for player retention and the long-term health of the game.

I'd rather just see the existing schematics in the Tailor line pulled and replaced with design elements. It would require

... got cut off there....

It would require rethinking and reworking the entire clothing system of the game, so I highly doubt that it would happen, but it would rocket off the end of the Cool scale.

Well... It all comes down to texture layering and city lag (Yes I'm bringing up city lag in regards to clothing patterns) The big concern here is that as it stands now, they have produced a fixed number of textures (the parts of the clothing that make up things like scales, like the little trangly things on the sleves of the Formfitting Shirt) and produce colors by biasing the vertex shading dome on the clothing. This means that for any given piece of clothing only a few, known in advance, textures are used. Now for those who are not familiar with graphics programming (i.e. most people) textures are little pictures that are 'mapped' to a 3d surface (like a shirt) to give it color and 'texture' (By texture in this case, I mean the dictionary definition of texture, not the programming one). Things like ribs, scales, and wrinkled cloth are all textures. Textures are large and their size is the whole reason for having more that 16 megs of video ram. Textures have color, but that color can be biased by using a material definition when defining the verticies of an object (the edges of the little 3d shapes). This allows them to color the item without having to create a new, memory intensive, texture.

Now in order to havea new 'pattern' (which in this case is directly synonomous with the dictionary defined word texture), there would have to be a new texture (the programming term this time) for each defined pattern. As well as a coordinate map for each pattern on each peice of clothing that can be 'patterned' (A coordinate map is more in depth than is needed here, but suffice to say, it's a lot of work).

This would require much more video ram, so since it would be unreasonable to load all of this into video memory at once (Imagine the havoc this would do to people with 32MB vid cards) they would have to dynamically load each 'pattern' as it was needed. This would cause 'hard disk churn' that would cause the game to lag every time someone with a new pattern came into the rendering field. Most of us would cry.

In any case, what this boils down to is that 'patterns' would be difficault to implement (If not impossible) at this time (In the future... Who knows? It could be like the SoL Expansion to EQ.)

However, interchangeable 'pieces' (sleeves, bottom components like dress bottoms, or pant legs to make a jump suit) are doable (With some work mind you. The people responsible for models and art would have a lot of work to do, but no where near as much as with patterns.) but would be a couple months away because of the design changes involved.

No matter how impractical it is, patterns would still be nice though.

So that you can charge even less for regular master items?

But seriously:

It should be easy to do. 256 basic item models, 256 pattern types, 2x256 colors. You can combine that in a 32bit value and the client can load the basic model and create the appropriate textures for it. Additionally 256 transparency patterns would add to it. Like this an elite2 clothing would require 5 bytes. Add another byte to allow more than 256 items (for future), it will take 6 bytes to describe the look of an item.

JimtheCactus are you by chance related to BobtheTomato? He posts around here. But I've never seen the two of you together.Hmmm.

Anway, I'm all for someone choosing the elite-elite profession. It would knock some master tailors out of the competition and move them up to make gobs of money for selling new schematics to tailors. A couple problems I forsee though:

How many tailors will want to purchase these new schematics if they themselves know that someday they will be able to design their own themselves? (maybe they would, I don't know that's why I'm asking).

Graphical glitches. I can just see all these different sleeves and buttons, and fabric, and torso pieces all looking great when a schematic is designed, but what will it look like when worn. I can see a lot of bad things happening since these outfits will have not been designed and tested by DEVS before hand. Then we'll have lot's of people screaming for the DEVS to fix this glitch.

I like the idea, if they can make it work.

True, you could store it in a long, but there's a problem: 256 Clothing Type * 256 Patterns = 256 New Pattern Textures (each being about 4 KB a piece) that have to be made and 65536 new texture coordinates that have to be set. SOE and Lucas Arts' art people would be busy for years making that kind of system. And using that technique (which assumes you're using alpha layering on the textures, if you didn't you'd have to make 65536 new TEXTURES *gag* or generate the composit on the fly, which could cause churn or at least a small amount of city lag) You've effectively doubled the GPU workload when rendering things with clothes. Storing the clothing details in the DB isn't really the problem. Storing that many unique textures in VRAM for rendering is the problem. 5 bytes to select which texture and vertex color to use, plus 2 (not 1 as is the current number) textures in memory (obviously twice the VRAM)per person. Like I said, not supremely likely if you plan to be able to walk in a city rather than giving you HD read head a good seek test.

Oh and BenjCaliman: No relation.

The art work isn't all that daunting. You could batch most ofit and just retouch.

The rendering issues are troubling, but I'm wondering if you could do it at the bottom of the priority stack, something like "Oh, here's a clothing texture. Slap the default on there and gimme the detail when you've got some spare cycles." So if you run past Joe Blowalker in town, the detail render never happens - just like in real life, you were too busy going from point A to point B to count his shirt buttons. If you're watching Joe boogie down in the cantina, the details fade in whenever there's enough free time to render them.

It's not a problem of detail quantity, it's a problem of detail layering. As it stands now, the texture being drawn is a 'texture' (Dictionary Term), big things like large folds and the scales on the formfitting undershirt are VERY noticable (the scales turn white to almost 50% grey, a massive, noticeable in passing feature) And the patterns (Which thinking about it, would also either have to be dynamically generated every time a character comes into the render field, or be nothing more than a monocromatic print that darkens the color, just like the current 'texture' texture does.) are just as noticeable (Would you miss the GIANT smiely face on alot of common shirts? And if it was delayed for rendering, when you stopped to look it would just 'pop' and become visible.) So a delayed rendering wouldn't work too well. By all of that is moot when speaking of GPU time. Any time two layers of texture have to be drawn in place of one, the GPU becomes burdened with twice the workload. But if your framerate isn't at monitor refresh or faster as it is, the software must provide fewer refreshes (i.e. FPS) in order to give the GPU the extra time it needs to render the image. Precaching thecomposite (i.e. precomputing the composite at the CPU and leaving the workload on the GPU the same) costs additional VRAM because for each player with a unique pair, the composite would have to be stored. And storing it on system RAM instead of VRAM is bad because it slows the GPU because of memory access wait delays. Raising the AGP apature (On systems that use AGP, if yours doesn't... Get a new system/vid card... ) would help there, but any time you have to go between card and system RAM it's slower because of bus limitations, and the additional bus demand would cut into the normal data flow which is overburdened as it is (at least on my 4X AGP card/bus).

Basically it would add a rather large amount of additional requirements to the base system (Raising required VRAM by probably 25%, only 25% instead of 100% because not everything drawn is a piece of layered clothing; and required GPU power for the same FPS, especially in places where lots of clothes have to be drawn, like in cities, etc.) And probably won't be a viable option for a couple of years.

(However, it would be awesome! Leather shirts, velvet dresses, dress blouses with a nice floural print... Ya... Awesome.)