PDA

View Full Version : optimizing images?


freez
25th July 2008, 03:10 PM
does anyone know of a quick tutorial in image optimizing? or anyway to optimize them for use in lightwave

bmckain
25th July 2008, 04:13 PM
I've been told the best method is "PNG" ;)

freez
25th July 2008, 04:48 PM
ahhhhhh the little light bulb just went on there, convert to PNG

thanks bob:tu:

mlynch
25th July 2008, 08:35 PM
PNGs are the best way to go (IMO). But when you are doing a project for print and you need lots of hi-res maps PNGs can start to bog down the scene. I've actually run into memory issues with this. In cases like this I'll use JPGs with minimal compression - I still get good looking maps when rendered, but the JPG compression reduces the file size and makes the scene more manageable.

But PNGs are my first choice. :)

BillS
25th July 2008, 08:36 PM
Easy enough to do. I normally save everything in .png but color is the only one that I leave as a 24 bit. Anything gray scale instead of using save as in the menu, go a little further down and use "save for web". The title may be a little different depending on what version of PS you have.

You should end up with a 2 up side by side. On the right select png 8. After a bit the image on the right will refresh. Take a look at the bottom of it and it will show you how many shades of grey are actually used in the image. Set the index amount to just higher.

11957

Once you load them in LW, that is where the big difference in memory will stand out. Here is the same image as a .jpg taking up 12 meg of ram compared to 4 for an indexed .png.

11958 11959

So it saved 8 meg. Doesn't sound like much but on a model with 20 or so maps and that done to all the diffuse, spec, gloss and bump. You can see where it would add up in a hurry.

freez
26th July 2008, 09:27 AM
that little light bulb just got a whole lot brighter :)

i have been using JPGs at best quality settings and ran into memory troubles, im getting near completion of my spitfire and started the maps when the problems started to occur,

thanks for explaining it to me :tu:

deg3D
26th July 2008, 10:44 AM
Not to brag, just to illustrate, I pretty much know Photoshop like the back of my hand having worked professionally in it since ver.1 and in the past having also taught it for many years, and I have run extensive tests on file formats configurations, esp. running all new ones when I got into 3D specifically for the purposes of 3D work, and I now save everything as PNG-8 256 colors. You can save color maps at PNG-24 if you like (which will result in a larger file size), but it doesn't really make that big of a difference IMO, maybe some for really up close shots.

Now, and most folks do not do this, as they are not website building-guys (which I am), rather than using the standard "Save As..." command, it is better to used the "Save for Web..." command to save out your images.

The "Save for Web..." command was designed to specifically to strip your files of all unneeded data.

Also, the nice thing about it is you can just save out your .psd workup files and you don't have to go through them and delete any superfluous data like channel masks, paths, etc. or anything else, as when using the "Save for Web..." command, Photoshop will automatically strip all unneeded data and leave you with the smallest file size file you can possibly achieve.

Hope this info helps all. :)

EDIT: Ah I now see Bill has expressed most of this info and with pics too. :) That'll teach me not to read the whole thread before posting.

Again though, I myself am not big on the use of PNG-24, but you can easily save out both PNG-8 and PNG-24 files to compare for yourself. ;)

deg

mlynch
26th July 2008, 05:54 PM
Now, and most folks do not do this, as they are not website building-guys (which I am), rather than using the standard "Save As..." command, it is better to used the "Save for Web..." command to save out your images.


...And this is where I fail! I've done a fair share of websites, but for some reason I feel I should only be using that option when actually working on a web site. ...excuse me while I find a wall to smack my head on.

CAClark
12th August 2008, 10:09 AM
I've always found simply converting the maps to 8bit before saving as PNG's. Do this for all your maps. Its quick and efficient. No messing about.

Typically even for colour maps, 8 bit will look no different, and the only time I go back to higher is for 32bit maps with alphas.

One thing worth look at also is PSD's as single layer PSD (8bit or higher) are quite efficient too, and make the transition between progress and final maps easier.

Cheers!

Gordon Robb
12th August 2008, 04:35 PM
Taking advice mainly from Mr Clark when I was doing the Bellflower. PNG-8 for everything works really well. No noticable difference when doing beauty renderes between 8 and 24. And the other piece of advice I think he gave me was, to have spec maps half the size resolution wise.

I do recall also something about having the resolution at set sizes 512x512, 1024x1024 etc...

billb
15th June 2009, 06:05 PM
Old thread, and a slight bit off topic, but is .PNG safe for use on the web now? The last time I looked, not all browsers would show them.

Quetzal
15th June 2009, 07:26 PM
I can' even read .png in windows from default, why should the browser be able?

on topic: is there really any diference between using PNG images and JPEGS for textures? can someone demonstrate a example? Nowadays I simply use stuff in photoshop format but I do it for comodity not quality but 99% of my textues are jpg

3D_CG
16th June 2009, 03:51 AM
What Mr. CAClark said .....

:)

Crook
16th June 2009, 05:03 AM
I can' even read .png in windows from default, why should the browser be able?

on topic: is there really any diference between using PNG images and JPEGS for textures? can someone demonstrate a example? Nowadays I simply use stuff in photoshop format but I do it for comodity not quality but 99% of my textues are jpg

I don't know if it is built into png, but by default it's not lossy like jpg, or so I believe, so jpg should always lose out for that point. For any artwork, lossy should be seen as unacceptable unless the situation demands it specifically. I'd like to have enough memory and power to cope with native layered images. I bet there would be some use for that in texturing, such as having all the maps in one file (diffuse, spec, normal etc etc)

Meurig
16th June 2009, 07:16 AM
PNG allows you to store high dynamic range data, as do a few other formats like PSD and EXR.

This is best explained with an example. If you get a jpeg picture of a fairly dark scene and raise the exposure on it, the jpeg has no extra depth to the data and so your pic is brightened universally. This isn't acting as true exposure in a camera would since you are just brightening it rather than getting any extra information from your pic.

If you were to raise the exposure on an equivalent 32 bit PNG version of the pic, it will bring out details that you couldn't see before. It's possible to make something visible that was entirely invisible before if the original HDR data is stored there.


For your average day to day CG work this probably isn't necessary and the general filesize and access speed efficiency of PNG is of more use, but the ability to properly expose extra data in these high-depth formats is what makes them really useful.

BillS
16th June 2009, 10:23 AM
on topic: is there really any diference between using PNG images and JPEGS for textures? can someone demonstrate a example? Nowadays I simply use stuff in photoshop format but I do it for comodity not quality but 99% of my textues are jpg

A visible difference? Not normally unless someone compressed the hell out of it as a jpg. The biggest difference is in memory usage. Just because a file has a smaller size doesn't mean that it has a lower impact on memory once its loaded. I've seen 2 meg png files take up 3 meg of ram where a 600k jpg balloons to 5 meg once its loaded and decompressed.

So not likely to visible differences but on big meshes with big maps it could make the difference between being able to render it or running out of memory.

Quetzal
17th June 2009, 03:18 PM
:-) So since I don't plan on changing exposure on textures and such. I'll keep using high quality, low compressed JPG's (if I had all my textures in *.raw or tiff I'd run out of disk space long ago)

deg3D
17th June 2009, 03:43 PM
Another consideration is keeping file dimensions at sizes divisible by 8. This keeps the decimal point from having to be calculated, thus speeds up render times.

With UVs one is best served by using the standard sizes of:

1/2K = 512 x 512
1K = 1024 x1024
2K = 2048 x 2048
3K = 3072 x 3072
4K = 4096 x 4096

Non-UV maps (never use 'em myself), just makes sure sizes are divisible by 8.

deg

Skruk
30th June 2009, 10:30 AM
Great info everyone. I'm taking it all in. Thanks!

Professor Moriarty
30th June 2009, 11:27 AM
(deleted... didn't notice how ancient this thread was!)

Tony Gardner
30th June 2009, 02:52 PM
(deleted... didn't notice how ancient this thread was!)

Ancient or not, I just learnt something by reading it and that's what it's all about ;)

BillS
30th June 2009, 05:43 PM
Necroposting CAN be a good thing :) :p