Noise???

Tom Reichner wrote in post #18287005
.

.

Not to mention (although I am mentioning), he makes some statements here in this thread that are inaccurate and misleading, and one statement that was actually non-sequitur ......... just how much should anyone trust the author of such inaccuracies?

.

Hmm. Sounds like questions being asked in other contexts around the world.

pknight wrote in post #18288819
Hmm. Sounds like questions being asked in other contexts around the world.

Or maybe I should cry.

Anyway, in agreement that it looks like what you can expect from that Sony shooting at that ISO. RAW would be better, but so it goes.

Also agree that seeing the original gives us context on whether the noise is unusual, as it depends on how many pixels we're actually seeing or if you've zoomed beyond 100%, etc. From an academic standpoint, I wouldn't try to say "that's normal" or "that's a problem" without having the unmodified original image to give proper context.

The fact that you have all the factors against you: shot at 6400, in JPEG, and are trying to lift details, will not help matters regarding noise.

You can certainly process it to reduce it but as mentioned in the previous post by TeamSpeed you will want to watch the faces and other details.

At 20x30 it won't be a stellar looking print, but it should be passable.

.

Colorblinded wrote in post #18288858
The fact that you have all the factors against you: shot at 6400, in JPEG, and are trying to lift details, will not help matters regarding noise.

And I think the biggest factor was that the image was underexposed (at least for the dark clothes), and that he had to then pull the darker areas up in post.

I can't help but to wonder how much less noise there would be if he had exposed 2/3 of a stop brighter. .A few tiny areas of highlights may have gotten clipped, but the tradeoff would have most likely been well worth it.

The second biggest factor affecting print quality is the fact that the image was shot so wide, with a huge amount of useless 'context' around the group of people. .So the end result will actually be as though he is printing the image at 40 inches or 44 inches, and then cutting a 30 by 20 inch section out of that big print......just another big hit on the image quality.

EDIT: After looking at the original again, I don't see any highlights that would have been clipped if the scene would have been shot at 2/3 of a stop brighter.....except for the area of glare coming off the wall above the people - but that probably won't even be in the print, as it will most likely be cropped out when the OP sets the desired framing for the print. So there really wouldn't have been any problem with giving this image a lot more light in the original exposure. .Shooting to the right will always result in less noise!

.

Tom Reichner wrote in post #18288861
.

And I think the biggest factor was that the image was underexposed (at least for the dark clothes), and that he had to then pull the darker areas up in post.

I can't help but to wonder how much less noise there would be if he had exposed 2/3 of a stop brighter. .A few tiny areas of highlights may have gotten clipped, but the tradeoff would have most likely been well worth it.

The second biggest factor affecting print quality is the fact that the image was shot so wide, with a huge amount of useless 'context' around the group of people. .So the end result will actually be as though he is printing the image at 40 inches or 44 inches, and then cutting a 30 by 20 inch section out of that big print......just another big hit on the image quality.

EDIT: After looking at the original again, I don't see any highlights that would have been clipped if the scene would have been shot at 2/3 of a stop brighter.....except for the area of glare coming off the wall above the people - but that probably won't even be in the print, as it will most likely be cropped out when the OP sets the desired framing for the print. So there really wouldn't have been any problem with giving this image a lot more light in the original exposure. .Shooting to the right will always result in less noise!

.

Yes, that's going to hurt it a lot. I didn't specify it but I alluded to it when mentioning "lift details"

This is far from "worst case scenario," at least the image was less than a stop underexposed, but it's not good!

Tom Reichner wrote in post #18288861
[So there really wouldn't have been any problem with giving this image a lot more light in the original exposure. .Shooting to the right will always result in less noise!

.

I shoot RAW nearly always (can't think of a time since 2003 that I haven't shot RAW in stills). The primary tone I pay attention to is the "brightest highlight that must retain detail."

I used to carry around a textured white plastic panel to use as a test target. I'd point the panel at the main light and raise exposure until I was just losing the panel texture on the LCD, then give it half a stop more (which I had determined by testing would put the RAW exposure right at the point of that highlight almost losing its texture).

So if the "brightest highlight that must retain detail" is exposed as much as possible without losing that detail...then shadows will have been given as much exposure as they can get as well. Looks too bright on the LCD, but works out just fine in RAW.

So I'm not just watching the highlights, but looking at particular highlights...and not really worrying about the shadows--they will get what they can get.

Remember also that if you need to be a little more aggressive than you would like with your NR, consider treating color more aggressively than luminance noise. In the crop posted earlier, it is apparent in the image that, in a YCbCr system, the Cb and Cr components exhibit large blocky compression artifact. These can be treated aggressively to remove low frequency, typically green and magenta, color blobs and restore more natural color. The higher frequency "grain" is primarily in the Y channel and can be treated less aggressively. The color is way more distracting in this image than the higher frequency Y channel artifact.

Finally, after noise reduction, consider adding back some "film grain" or whatever it may be called in your specific application. This will usually restore a more natural look to your image, breaking up the plasticky look of aggressive noise reduction and potentially adding some acutance (perceived sharpness) back into your image.

Attached is the YCbCr channel representation of the crop. You can see all of the JPEG blocks easily in the Cr and Cb channels.

kirk

Attached is the crop (original) and the treated crop - both have had a curve applied to boost the overall brightness. The treated image has had noise reduction applied, then some grain added and also some local contrast added. It is difficult to tell if this treatment would be sufficient to print this image.

kirk

Again, I appreciate everyone's comments and time.
I started again from scratch, did what I could with keeping as much detail as I could, and sent it off to my printer to have a 20x30 metal done for my parents.
It is what it is.
There is a list of things (RAW, compose better, watch exposure/lighting, for starters) I could have done better, and will hopefully address these next time.

Some photos are treasured because they demonstrate good technical craftsmanship. Some photos are treasured because they embody artistic inspiration. And some photos have neither, but they are no less treasured because they make special moments eternal.

tzalman wrote in post #18290053
And some photos have neither, but they are no less treasured because they make special moments eternal.

That's a romantic, flowery notion. In more cases, some photos have neither, but they are valued only because there's no alternative to the record.

Or the photos, however poorly composed or captured, provide blackmail material or tabloid material.... Very valuable indeed!