MicrostockGroup Sponsors


Author Topic: Downsampling Images  (Read 5221 times)

0 Members and 1 Guest are viewing this topic.

« on: March 03, 2008, 06:11 »
0
Hello everyone,

I just wonder - do you usually downsample large images (say 12Mp) before uploading them to e.g. Shutterstock?
If you do, then I would like to know whether it is better in your opinion to change the size of the image in camera RAW before opening the picture in PS, or after all changes made in PS using the Image Size - Resample - Bicubic?
Thank you for your help.

Have a very nice day.


« Reply #1 on: March 03, 2008, 06:27 »
0
after all changes made in PS using the Image Size - Resample - Bicubic?
I do all the postprocessing on the 16-bit TIFF from the raw, and sometimes with two developments for problematic highlights. When all is done, I save that 16-bit TIFF as the master copy since it's lossless and it's the vault of all my postprocessing time. The raw isn't.

Resizing and converting to 8-bit JPG is always the final step and when changes have to be made afterward, I always start from the saved TIFF again.

Resizing I do in PS > bicubic sharper, and on the 16-bit TIFF. Converting to JPG and save as JPG (quality 12) is always the last thing.

For SS, I downsize from 10 to 6MPx.

« Reply #2 on: March 03, 2008, 06:44 »
0
Thank you very much - I usually do it exactly the same way, but a colleague of mine insists on resampling the image in camera raw from 10 to 6 Mp. Then he sometimes does the downsampling AGAIN in PS, which I think is a nonsense. But he keeps telling me he gets much sharper images that way.

« Reply #3 on: March 03, 2008, 07:40 »
0
Well sometimes people forget to turn the in-cam sharpener off. In the Nikon RAW developer, I think there is some sharpening on by default. That might have misguided your friend. Just turn it all off. A customer can easily sharpen himself, but he can't undo your sharpening. You can make omelet out of eggs, but not eggs out of omelet.
I think also you can get some distortion when downsizing sharpened images. Downsizing sharpens anyways when choosing the bicubic > sharper algorithm.

The story is not complete. I found a trick on a tutorial site to oversharpen 4x upsized shots in the Lab colors lightness channel alone, then resize back to normal. It avoids halos. Blend this layer at just 10-30% opacity with the original one and you will get that little bit of extra crispiness we all love, but I always keep the original unsharpened file. Sharpening can introduce really bad artifacts when you look at 100%.

« Reply #4 on: March 03, 2008, 07:49 »
0
Well, I do not know about him this very moment, but I have just had a look at my camera raw and there is some default setting for sharpening 25 and colour noise reduction 25. Shall I just turn it down to zero?
BTW. in the very rare case I do some sharpening myself, I always use only the Lab Colours and the L channel only. And then the "unsharp mask".
But I like the trick with the blending new layer you spoke about. I am going to try it right now.
 :)

« Reply #5 on: March 03, 2008, 17:08 »
0
For SS I downsize from 17MP to 4MP using the Bicubic Smoother option. Given the general sharpness of the originals and massive amount of downsizing, I find the results more pleasing to the eye using this method.

« Reply #6 on: March 03, 2008, 20:51 »
0
sharply_done,

I think you may have mis-spoke when you referred to using bicubic smoother. That is the method used when uprezzing. (up sizing an image)
To down size an image, it is generally the bicubic SHARPER that we use. I know that you just got them reversed, and meant it
the other way around.

I just thought I would jump in before anyone newbie actually used the smoother to downsize.

Sorry for the interjection,
The MIZ

« Reply #7 on: March 03, 2008, 21:08 »
0
Nope, I use 'bicubic smoother' to downsize. My images are already sharp enough that I don't need to sharpen them further using 'bicubic sharper', and this option also smooths out problem areas a bit more.

As I said before: I think things look better when I use this option, and I never get images rejected for being too soft.

« Reply #8 on: March 03, 2008, 21:23 »
0
sharply_done.

I'm glad that this method works for you and your images.

Just so you are aware "Bicubic Smoother is a new interpolation method specifically designed for upsampling. As its name suggests, it gives a smoother result that handles subsequent sharpening better than Bicubic sampling."

I will include this link http://www.peachpit.com/articles/article.aspx?p=433340&seqNum=5
so that everyone can benefit from what the experts have to say on the subject of resampling.

The MIZ

« Reply #9 on: March 03, 2008, 23:29 »
0
Ordinarily I don't reduce the size of any images submitted to any agency (including SS).  I submit full resolution.

Very occasionally I will reduce the size if I feel that quality or editing is 'borderline' and in these instances I use bicubic or bicubic smoother (like sharply) to reduce to 90% or 80%of original.  I've found that 'smoother' gives a more even tonal result than 'sharper' when downsizing.

vonkara

« Reply #10 on: March 03, 2008, 23:52 »
0


Very occasionally I will reduce the size if I feel that quality or editing is 'borderline'
Hatman, I wasn't know that you was producing borderline files ;) And congrats for your brand new D300, Hope you enjoy it

« Reply #11 on: March 04, 2008, 03:12 »
0
Thank you all very much. So in general - you all agreed downsampling (if any) should be done as the very last thing in PS - using bicubic interpolation (or b.sharper or smoother - whichever will look better and work better for me) and NOT in camera raw before editing.
Thank you very much and have a very nice day.

cphoto

  • CreativeShot.com
« Reply #12 on: March 04, 2008, 15:20 »
0
Nope, I use 'bicubic smoother' to downsize. My images are already sharp enough that I don't need to sharpen them further using 'bicubic sharper', and this option also smooths out problem areas a bit more.

As I said before: I think things look better when I use this option, and I never get images rejected for being too soft.

I go from 21MP to 2.5 - 4MP max with SS, and up to 8MP with FT, and I use the same method.  I agree that bicubic smoother seems to be working better, at least for me.  I think it helps to remove noise and other problems in the picture.


 

Related Topics

  Subject / Started by Replies Last post
157 Replies
47188 Views
Last post December 28, 2008, 20:03
by Tuilay
5 Replies
3423 Views
Last post February 16, 2007, 06:01
by sharply_done
2,000,000 images

Started by Dreamstime News Microstock News

0 Replies
2219 Views
Last post November 20, 2007, 06:57
by Dreamstime News
4 Replies
3351 Views
Last post January 17, 2008, 16:50
by ZMB
2 Replies
5254 Views
Last post November 28, 2008, 12:18
by Peter

Sponsors

Mega Bundle of 5,900+ Professional Lightroom Presets

Microstock Poll Results

Sponsors