1024px to 512px

I used search but couldn't find, so here is a question for you texture-experts out there!

Depending on what project I'm working on - my original psd's sometimes are 512px and sometimes 1024px (PhotoShop)
For my skins I use 512 because I like to have my pixels in total control - but for clothes 1024px.
I always use TGA's with an alpha-channel for transparency.

My files usually have a trillion layers... And when working in 1024 I find it a little time-consuming (tick, tick, tick) to shrink the whole psd-file to 512 before I save and upload the textures to SL. Not to mention forgetting to turn the original back to 1024 and not having a copy (stupid, but happened, grrr).

So...

I could save the texture as a TGA (with Alpha) - then open it up again, shrink it to 512

or

save the texture and the alpha-channel in separate files as 24's, then open them and create a new psd (with the alpha) and save as 32's...

I've done all of them and to me it seems to come out the same... But that might be a 'naked eye' thing.

Will there be any difference in quality between doing it one of these three ways?
Is there a better way? How do you do your shrinking?

This is what I do.

Also, to save time, I've created an action set with one-click resizings for all SL-compatible sizes. I'd recommend everyone do this. It takes just a few minutes to set up, and then it saves you several clicks on every file forever after.


That's pretty unnecessarily complicated.


The second two ways will come out exactly the same, if I'm reading you right. Either way, you're simply resizing a flat, uncompressed image. You're just getting to it by two different sets of steps.

The first one, resizing the original PSD, is where things could get wonky. First, as you mentioned, it's easy to forget to undo, and then you're screwed. Second, if you're using any parametrics such as layer blending effects, vector strokes, etc., then resizing the image can potentially change things. The relative proportions of all the various parameters might not perfectly survive the resizing.

This is how I do it, to preserve effects to their relativity and such as well.

Save PSD.
Flatten, Resize.
Save TGA.
Undo or load previous PSD.

Thanks Chosen!

The more 'complicated method' I sometimes find easy when I'm done with the 'layer-part' of a project and want to add other stuff like different coulors, different buttons or such. Nice to know both work the same way!

As far as I know: You shouldn't flatten... Layers should be intact!

For the: Resize/Saving/Undo - that is what I think most ppl do. But as Chosen said it might not be the best method...

Correct

There is absolutely no reason to flatten the image whatsoever for uploads to SL.

Flattening can result in the loss of all the original artwork and additional hard disk space being used along with additional workload.

For example I design some textures in one file and in that file there could be lots of different layers that make up several textures. I just resize that file, select the layers I want that make the texture and save as tga. then with the same file open I select my other layers that make the other texture and save as tga. alternatively I might have already merged the necessary layers that make each texture and selecting each one to save as tga. I don't always merge as this is not always necessary and results in having larger files taking up more disk space.

Now if I were to use the flatten option I would have to open the same psd file over and over again after flattening each image just to save as a tga, a total waist of time and there is no difference whatsoever to the finished image on upload.

So my advice is get out of the habit of of flattening, the worse you can do using your psd files is select yes to save the file after you save as tga. This just means the file will be 512x512 which was the upload size so selecting no and not saving means my file is in its original form and size with no loss of layers etc as can happen with flattening.



Olila, when you shrink the image to 512 why should you have to change its size back?
If you close the file and select no to save after saving as tga your original file will be well, in its original format.
Only if you select yes to save then your original PSD file will be overwritten with the new image size.

Egads. I thought there was a problem with my workflow. I'm glad I'm not the only one with this problem. I thought there was just something I had yet to learn. Do I do it the way I've always done it and get it done now, or do I spend time now to see if there's a better way?

I've just always merged layers (because PS messes up the edges of a layered image when you re-size), then re-size, save a copy, then undo the resize and layer merge with the history window. Makes me nervous every time.

I agree. I always save the PSD file at its original scale, then downsize to create the TGA at 512 x and close the file. So long as I don't make the mistake of answering "Yes" when PS asks if I want to change the PSD file, it's safe at its original size.



Hmmm... I've never had the edges of a layered file get fouled while resizing. Maybe I just don't know what to look for. I'd rather not have to merge and unmerge layers unless there's a good reason to. Do you have a photo of "messed up" edges?

Well, that is one way of doing the same thing but most of the time I THINK my uploaded TGA is my 'final cut ' - but often I decide to make some changes and then I 'undo' the shrinking and keep on working... If I'm satisfied with the upload I just close and choose 'No Save'.



Hmmm! Never noticed a problem with that... Maybe it depends on what techniques one uses...



You don't have to spend time...
Regarding what Chosen and Larrie said (tossing in my own experience) you should:

* Merge the layers you want to merge while working on your project
* Keep the these layers without flattening and also keep your working-size
* Save the file as TGA
* Open the saved TGA-file and resize it.

That is the way I usually do and it has always worked great for me.
But I was curios if the true experts knew something I didn't.
In my opinion it's the easiest way! Hurray!

Edit: Rolig Hmmm'ed the same thing while I wrote my post...

I just make a duplicate image, with layers merged, resize and save as a targa, and have my photoshop file separate.

I've seen this. Try this...

Create a new 128x128 image, white background. Add a layer and fill it with black. Re-size the image to 64x64. Zoom way in on the edge. Looks black, right?

Use the eyedropper to sample the edge, and take a look at the color picker. It's not black. It's 13-13-13 or 15-15-15 or so. Now, hide the background layer. The edge looks semi-transparent now, and if you sample it with the eyedropper it's black. So, it apparently is partially transparent, and against a white background gives you a very dark grey. This happens with all the bicubic resizing algorithms in Photoshop.

Now un-do your re-size, merge layers and re-size again. Zoom in on the edge and it's black.

This isn't very noticeable unless you are moving or copying around pieces of your image. Keep the edge on the edge and it's not usually a problem, but move that edge against another background and it becomes noticeable. A grey border shows more prominently when you position it against a larger, black background. It happens with any color, but I notice it a lot because I work with black a lot, and I re-composite images into other images a lot.

I've seen it too, and it's as annoying as Hell.

Wow. I see what you mean. I would never have thought to do that experiment.



Thanks, Anya. I don't tend to move and recomposite image chunks as much as you seem to, so I have been blissfully ignorant. It's something more to add to the watch list .....

If I understand how image resizing works (which I may not ><, I would recomend resizing the image to 512 px before saving it as a tga file.

Why?

Because when you save to the .tga format, it automatically reduces the resolution to 72 dpi. So, if you're working in 150 dpi (or higher), you'll be losing a lot of detail by converting the image to .tga... then you'll lose even more detail by sizing the .tga file down to 512 px. That's quite a bit of data loss, so I guess it's up to the individual of how important image quality is to them.

Someone please correct me if I'm wrong about this.

You're wrong about this.

DPI makes absolutely no difference whatsoever unless you are sending your file to a printer. You can change that number to anything you like and it won't affect the appearance of your image on a monitor one little bit. It's simply a scaling factor to let a printer know how many pixels to cram onto a sq inch of paper. Monitors don't do anything with that number. After all, the number of "dots" that are crammed into a square inch of your screen depends on how much you zoom in on the image and what the resolution of your monitor happens to be.

LMAO

If anyone is deciding to catalogue their work and print their textures prior to upload to SL please consider the above and do not save as tga prior to printing.


Okay, sorry but you are wrong.

Correct

The default resolution in PS is 72dpi and this does not change at all when you resize your image.
Anyone working on texture design using more than 72dpi is waisting mega amounts of HD space for something that will look exactly the same in SL if designed at 1200dpi or 72dpi.

Not necessarily. If an image has a fixed amount of pixels in it, it doesn't matter how many pixels happen to fit into whatever is defined as an "inch". For example, an uncompressed single-layer RGB image at 1024x1024 pixels will always take precisely 1MB of storage space. The dpi (or ppi) setting cannot change that. The number of pixels times the bits per pixel equals the size of the file, period.

The label could say 72 dpi or 7 million dpi, and it won't change the file size. All it will do is change the physical size at which the image prints, if it's ever printed.

It's only when you start using dpi settings to resample the image that things change. When you do that, you start adding or subtracting pixels from the image. So of course the file size increases or decreases accordingly as well. But if you're not resampling by it, then "dpi" is just an arbitrary label, completely meaningless for texturing purposes.

What's relevant here is that the dpi change Skuz mentioned is NOT a resampling. It's simply a re-labeling. It doesn't affect the image's file size or on-screen appearance in any way, shape, or form.



I realize you already understand this, Larrie. I just thought the way you phrased it might be potentially misleading, so I figured a little further clarification was in order.

I just experimented with resizing and changing the resolution of copies from the same photoshop file and you guys are exactly right! There was no change in image quality at all, so I apologize for my obviously flawed theory... I guess spending 10 years in print media really corrupts your thought process when it comes to working in a strictly digital enviornment.

Again, sorry for the confusion. X}

Thanks Chosen

You know I am not always the best one for explaining in so much detail so clarification is always appreciatted.

Ahh! Really weard..

But now I remember seing it before! Thats the reason I usually use 'crop canvas' when making haitextures! On most of my other files the blurry edges doesnt matter...

Thanks for the reminder!