New upload prices: which is better?

x: texture size
y: cost
1-) Constant size/price rate (k=x/y):
256X256 (64K) = L$10
512X512 (256K) = L$40
1024X1024 (1024K:1M) = L$160

2-) Parabolic size/price rate (k=sqrt(x) / y ):

256X256 (64K) = L$10
512X512 (256K) = L$20
1024X1024 (1024K:1M) = L$40

3-) Leave it as it is.

I'd go with the first option, but make 512*512 the base size for a 10L fee.

Actually I'm more for parabolic (in the other direction) where you pay more per kb for a large file.

So

256x256 at L$5 (not 2.5 ... to account for overhead)
512x512 at L$10 (base)
1024x1024 at L$160

My rationale behind the higher per KB rate for larger files is that large texture files slow down our loading time when we see them. The pricing scheme above will make 512 the de facto standard with 256 used by those who want to save on upload costs.

The pricing scheme above will force people to do better 512's, use 256's when they can, and use 1024s only when absolutely necessary.

Exponential would be good.

It would also give the rich the ability to create higher quality builds than everyone else. Not so good, surely?

Just textures... and the best textures in the world won't help if you can't build to save your life. Counter-example being Maxx Monde, who, though renowned for being one of SLs greatest builders, went for over 2 years without using any textures... just blank and shiny.

Higher quality very rarely has to do with utilitizing huge textures, and if it'd make people think twice before dropping these 1k files on everything then it'd be a nice thing... however, since four 512 x 512 textures cover the same area 1024 one does, suspect it'd just result in people using 4x as many prims just to walk around the upload costs. :/

However, those 2 years were in the past, before SL had some of the beautifully textured builds it has now.

Some are still better. Texturing is not building. It's a seperate skill. Makes for some nice views of course, no denying, but being able to bake a texture does not a builder make.

Using 1024x1024 does not always translate to higher quality. But it always translates to higher loading times.

Higher quality for me is a balance of detail level, prim count, and loading times. The faster something loads, the sooner people will enjoy it.

The problem is that some people use 1024s as a means to supposedly make their stuff look good, but don't succeed. It's either because the texture is bad to begin with, or the texture would still have worked at 512.

at 512, you have 51.2 pixels per meter (if it's a 10m object). That's pretty decent. And considering that the best things I've seen do not put much detail on 10m prims, 512 pixels along one dimension is plenty. A 1m object with a 512 texture has 512 pixels per meter.

If an exponential scale means that only the rich and adventurous will use 1kx1k textures ... then so be it. That translates to more Linden$ sinking and that's good for the economy.

If it means that less people will use 1kx1k textures, GOOD ... it means less lag.

As a relative (and relatively poor) newbie, I really don't want the prices changed.

This makes it harder for newbies to learn how to make items that require textures to be uploaded (can't tell you how many times I've uploaded something that didn't turn out exactly right even after previewing).

In my case particularly, magazines for the ThinC printing press get a lot more expensive. Each page costs 10L now; for a 16 page magazine, that is 170-180L including frontcover, backcover, spine, and sides. (And 16 pages is quite small for a magazine, others go up to 64 pages or more.)

Under system 1, it could cost up to 16 times that. 2880 for a 16-page magazine distributed free? That's a lot.

Leave it as it is, but restrict uploads to 512 x 512 or less.

There is a companion proposal to this from sometime back that if prices are changed, there will be a grace period for loading. The charge applies only after the file is confirmed If you don't reject the file within 30 minutes you will be charged. That means that you upload, use it for a while and check it out. If it's wrong, reject it and no charge applies.

I still think they should go by file-size, the bigger (more detailed) the file, the higher the upload cost. This way a large texture that compresses well (e.g a photo or something) doesn't cost loads, since it transfers much more quickly and thus uses less bandwidth.
However, a large file, with large (e.g lossless) detail will cost you much more.

I'd do it exponentially/parabolic, with the base-line being 256 x 256 at L$15, 512 x 512 being around L$25 and higher being quickly more.

Having 512 x 512 as the base, while it would appease all the clothing creators who will moan about it, would defeat the purpose IMO, as 512 x 512 for L$10 is currently a steal.

It would be far more efficient if SL could just use automatic downsampling though

Ehh?

if L$10 per upload is 'currently a steal' and 'defeats the purpose' then what exactly _is_ that purpose, then?

at L$ 10 per transfer the upload charges still manage to create largest currency sink (over 6.5 mil burnt there montlhy) ... there's absolutely no say that increasing per-transfer charge on everything is going to increase amount of money wasted. The effect might be exactly opposite, i.e. people spending the same overall and consequently contributing _less_ content.

To be honest, I hate slow loading sims just as much as the next guy, there is not a doubt that it lessens the experience. It is also true that having great textures does not make a horrible building become a masterpeice...

However...I am not so sure that effectively limiting the size of uploads, which is what I presume the purpose is (since SURELY it is not a program to punish the poor) would be the best way to INCREASE loading speeds.

It seems to me that the best way to increase loading speeds in an area is to keep pressure on LL to increase the efficiency of their platform. Measurable results from a platform level correction would be seen immediately across SL as opposed to limited increases in New Sims, or currently underdeveloped areas where new building is waiting to take place. Remember, the large files that are causing slow loading times are ALREADY here....they are not going to go away with attempts to limit the size of NEW files.

As a result, I believe that the overall increase in loading efficiency from this plan would be insignificant, as far as the OVERALL grid goes.

I think more of a solution to this comes from the ability to store data from the grid on the home system, reducing traffic and delays from the Server trying to shove all the details down to the user.

So, for the purpose of increasing loading speeds, I would have to vote to leave it as it is. But, it does make a great money sink. <grin> I might support it if THAT was the purpose.

Maybe increase the cost of textures with more pixels than a 512x512 to L$20-40, if you absolutely must. While there are legitimate uses for textures that big, they're generally few and far between.

I'm against increases for normally sized textures, though. Newbies already (probably partly thanks to this forum) start with no money. The harder you make it for them to experience SL's various features, the less likely they'll stick around and become paying customers.

I think the uploader part of the client should automatically calculate and rescale the image for optimum speed and the cost should be L$10

I don't think increasing the price of uploading texture will solve any 'loading' problems. As stated earlier, some big textures are already out there. The grid needs to be upgraded by Linden labs and not on the backs of its paying and free players.

What i do in game costs me a lot...i mean ALOT of L$ in texture uploads and I do use 1024x1024 because it alows me to give my work more richness and detail. Making my textures in 512x512 actually really does decrease the detail - i have tried it both ways. I make tattoos, really nices ones - not your solid black tribal - those are fine at 512x512. All i can say is any increase would have to be reflected in my in game prices to other players.

This loading problem is a Linden lab problem...lets leave it to them to upgrade the grids memory and loading aspects.

parabolic

Niether. Keep the current costs.

Wow...

I'm shocked people actually use 1024x1024 images...

I feel guilty if I have to use a 512X512.

I like the OP's #2 suggestion however.

You guys do realize that textures don't have to be square, right, as long as both sides are powers of two?

The only points of altering upload fees are to increase SINKS and encourage intelligent design. If you're doing that setting 512 as the base size is purely pointless. You're still using oversized textures on items small enough to not NEED the higher resolution which wastes bandwidth, vram and cache space. So choose values that both INCREASE the overall fees paid, encourages people to CHOOSE a size based on NEED (not whim) and all around actually has an effect. Here:

up to 32px * 32px area 1$L
up to 64px * 64px area 5$L
up to 128px * 128px area 10$L
up to 256px * 256px area 15$L
up to 512px * 512px area 20$L
up to 1024px * 1024px area 25$L
anything larger 50$L

Screen Shots
low res 5$L
hi res 50$L

Automatic upload of screenshots I leave at a reduced rate for normal screenies to encourage socialization. Hi-Rez mode screenies are ludicrously large so I dont offer them any immunity.

Now if your intentions are less to help the economy and ONLY to encourage intelligent texture use a set of prices of

up to 128px * 128px area 5$L
up to 256px * 256px area 10$L
up to 512px * 512px area 15$L
up to 1024px * 1024px area 25$L
anything larger 50$L

would suffice. values i've supported before.