noise reduction before or after enlarging ?

BS
Posted By
boarder_s_paradise
Oct 24, 2008
Views
858
Replies
15
Status
Closed
Hi. I’d be interested in your opinion:

If the circumstances force me to work on images with noise which have to be enlarged, and if I use noise reduction based on noise-pattern-recognition, is it better to carry out the noise reduction before or after the enlargement (and in case the answer is after, which resampling method would resample the image (and thus the noise) best for subsequent noise reduction?)

And what’s the logical rationale behind the answer ?

Thanks a lot!

PS: I think the answer will be the same for any type of noise, but if there are significant differences between noise types, I’d be interested in that, too.

How to Improve Photoshop Performance

Learn how to optimize Photoshop for maximum speed, troubleshoot common issues, and keep your projects organized so that you can work faster than ever before!

BS
boarder_s_paradise
Oct 24, 2008
PS: Does the answer change if the enlargement significantly alters the aspect ratio, let’s say X’:Y’ = 4*X : 2*Y
?
MV
Mathias_Vejerslev
Oct 25, 2008
Before, because the noise is expected to be about the size of a single pixel.
BS
boarder_s_paradise
Oct 25, 2008
Thank you!
BS
boarder_s_paradise
Oct 25, 2008
Trez Hane, of course I had tried it before but I never saw enough difference between both ways to determine which one would better quality-wise.
So I wanted to know as a matter of principle. (For example if I know that one method always yields poorer results, I’ll never try again).

And besides, even if I had seen from experience which way has better results, I guess I’d still have posted to understand the reason for it.
TH
Trez_Hane
Oct 25, 2008
Just as a further thought, why don’t you try it out for yourself both ways using small samples? After all, it is you who must be satisfied, and it would only take you a matter of minutes. And you would be working with the type of images that matter to you.
BL
Bill_Lamp
Nov 5, 2008
NoiseNinja instructions say to apply it as early as possible in the work flow. Right after the file leaves ACR is where I use it [CS-3 user]

Bill
RK
Rob_Keijzer
Nov 5, 2008
It should even be done before lens correction (abberations, vignetting and barrel/cushion distortions)

Rob
F
Freeagent
Nov 5, 2008
I had some seriously noisy images that I tried different approaches on. Best result by far was to save as TIFF and do noise reduction in ACR before opening into Photoshop. ACR is great for this, especially in combination with initial sharpening with the detail slider set to 0.

Another useful trick is to do noise reduction inside a reverse luminosity mask, leaving the highlights more or less unaffected.
JM
J_Maloney
Nov 5, 2008
Another useful trick is to do noise reduction inside an inverse luminosity mask, leaving the highlights more or less unaffected.

Can you explain this Freeagent?
JM
J_Maloney
Nov 5, 2008
Ok, I understand. 🙂
JJ
John Joslin
Nov 5, 2008
Hey! Is this telepathy? 8o
BS
boarder_s_paradise
Nov 5, 2008
Thanks for all your comments.

I also thought that NR should be applied as soon as possible, leaving the noise patterns intact. On the other hand I thought that enlarging also enlarges residual noise, so maybe the bottom line is to do a regular NR initially and then, after enlarging, a second weak/careful one on parts of the image if need be.

@Freeagent: very good tip the inverse luminosity mask! As for ACR: From my personal experience, ACR noise reduction is very effective on RAW material, but on all other sources it far lags behind noise-pattern-recognition methods.
F
Freeagent
Nov 6, 2008
Hey! Is this telepathy?

Yes, I figured it would be simpler to just beam it over 😉

Anyway, if that was unclear to anyone else, it means ctrl-clicking on the RGB channel (or Luminosity in Lab for that matter) to load it as a selection, invert it so the dark areas are more selected, and then noise reduce (reduct?).

The point is that noise is a bigger problem in the shadow/less exposed areas, while the highlights may in fact be OK. And that’s where most of the important detail is.

I’ve found ACR to do well on TIFFs and JPEGs as well, but of course there’s no formula. The main advantage is to combine noise reduction with the "0-detail-slider" sharpening (and perhaps mask) to keep as much detail as possible while still getting rid of most of the noise. And you can check the combination in real-time while you play with the sliders.
JM
J_Maloney
Nov 6, 2008
You explained it much better in my head. 😉
F
Freeagent
Nov 6, 2008
LOL!

You know how it is…you try to explain but the words get in the way…

MacBook Pro 16” Mockups 🔥

– in 4 materials (clay versions included)

– 12 scenes

– 48 MacBook Pro 16″ mockups

– 6000 x 4500 px

Related Discussion Topics

Nice and short text about related topics in discussion sections