Grain or Noise

[chris_fernando]

<p>I'm trying to determine if this is grain or scanner noise. My lab is telling me that this is film grain and typical despite being a 4x5 transparency (Velvia 50) being blown up to 24"x30". I'm a bit skeptical, however. Is this film grain or scanner noise? It's most prominent in the darker (greenish) water up at the top and looks like a mottling or pixelating-type effect but with the randomness of grain. Any help would be most appreciated. Thanks, in advance, once again.</p><div></div>

[janne_moren]

<p>I don't see anything amiss in the image. Could you be a bit more specific, or post a crop where it's clearer?</p>

[chris_fernando]

<p>Unfortunately the combination of web and JPEG compression is smoothening it out in this shot. It's quite evident in the 7MB TIFF though. I'll try to describe it but I haven't quite seen anything like it - the best way I can describe it is "noise" sort of like the video compression you seen when your blacks "creep" in the video realm; but not blocky or pixellated, it's more like a patina, which may be grain like, but it seems too artificial (read: digital) to be organic or film based.<br>

Sorry if I'm being vague, it's somewhat hard to describe. It's most evident in the transition areas between light and dark in the darker area of the water above and on the shadow side of the rocks in the water in the foreground area; what looks like a spattering of dots making up areas, or blobs, of different tonalities/densities. My first thought was some type of compression artifact as (I don't think) film grain affects shadow areas to the degree that this is, but I know underexposed areas of film grain tend to be more evident. </p>

 

[chris_fernando]

<p>I was able to attach the TIFF version, where it's clearly evident. Thanks, all for any insight!</p>

[janne_moren]

<p>If you mean the shadow side of the rocks, that looks like the scanner is blocking up. The black is darker than the scanner can pick up, and the brighter pixels is the lowest non-black. Then, as the image is brightened this point is brightened too - but the black stays black, so the transition becomes really visible. The best way to deal with it for me has been to make a small "toe" on the image curve in postprocessing so these areas get a decently smooth transition (at the cost of darkening these areas a bit of course).</p>

<p>Another thing: you seem to have a fair amount of sharpening active (there's halos around your dust spots). That will accentuate this (it brightens the pixellated transition), and increase the noise in your image too. It's generally not a good ida to have any sharpening unless you're going to print directly from the scan.</p>

<p>Edit: Are you by any chance scanning at 24 bit color and brightening the scanned image afterwards? The banding effect in the far background looks kind of like when you do big global brightness adjustments to an 8-bit per channel file. Editing in 8 bits is fine, as long as you've set the overall brightness and color of the image in 16 bit, either in the scanning software itself or in some external tool (I use UFRaw for my scans).</p>

 

[chris_fernando]

<p>Thanks Janne and Lee. I guess this is what you run into when you rely on a lab to do the work for you, albeit at a price I thought would not leave so many unanswered questions.<br>

Janne, sorry for the ignorance but when you mention not applying sharpening when printing directly from the scan, I'm a bit confused - what else would you be printing from if not the scan? To answer the question, though, yes I will be printing directly from the scan, in that the lab is scanning this and printing directly from the file, after applying some color correction and some sharpening (a no-no, perhaps?). Excuse my ignorance.<br>

Lee, they claim they are going to "clean up" the dust/dirt and sharpening anomalies, but if I don't see what I'm beginning to think I should be seeing on that final print when I pick it up, they may have one more for the trash heap.</p>

[chris_fernando]

<p>I sent you a private e-mail regarding possibly using your scans to show the lab what is possible. Thank you.</p>

[chris_fernando]

<p>Les, your 4000 dpi Coolscan file has got the same banding/noise issues (In the dark water area in front of the twig that is sticking out of the water, just right of the center of the frame, in the background of your photo) that I see on my scan. The banding/noise is nowhere to be found on the 8000 dpi Imacon scan. Is the improvement a direct result of the increased resolution?</p>

[chris_fernando]

<p>Thanks, Les. I see what you mean by the difference versus what's in the response versus what's on your site. The reason I pointed out the anomalies and what I assumed to be the scanner/resolution combo is because the anomalies that I am seeing at 150% on my monitor are as close as anything I have been able to find, to the anomalies on my print. I am basically seeing almost exaclty what I am seeing in the shadow areas of your Coolscan file, but my concern is that I am seeing them at about 3 to 5 feet away on a test strip print that is a portion of my final print size (i.e. 30 inches high by 3 inches wide - a strip to approve basically). From what I am reading that's not organic grain, but more like grain via a sampling error (?). I'd appreciate your thoughts on what that might be.<br>

The store is using a Scitex EverSmart Pro scanner. They are in San Diego, CA.</p>

 

[janne_moren]

<p>It's not resolution, but the ability of the scanner to "see" in those darkest areas. If aq negative is too dense the scanner sees only blackness. Yes, the Coolscan scan shows a bit of that. Slide film with its high overall contrast is fairly difficult to scan well for this reason.</p>

<p>Lee, what I meant by "print directly from the scan" was exactly that: scan, then send the scanned image to the printer without any postprocessing in between. If you do that then sharpening as part of the scanning makes sense. But if you intend to postprocess the image then it's much preferable to have an unsharpened scan, and apply sharpening as the last step in that process instead.</p>

<p> </p>

[bryan_wolf]

<p>Chris,<br>

You've probably already got this sorted, but here are my thoughts.<br>

I own an eversmart Jazz+ scanner so I have some familiarity with the Creos. I think you are seeing a combination of a few things here. The eversmart pro is an 8 bit scanner(output;14 bit internal) so there will likely be some posterization, especially if there are major changes done to the output file rather than in the oxygen scan software.. Also the lower end eversmarts <em>can</em> exhibit noise if pushed too hard. Is that slide under exposed? My scans of velvia can and often do, look like this, if the slide is very dark and oversharpened. <br>

<br />My Opinion is that what you see is definitely not grain, but likely a combination of posterization, noise, and grain aliasing. That said an Eversmart Pro should easily make you a great 20x30 from 4x5. Assuming the optical path is clean, they deliver 95%+ of their advertised resolution and D-range/max.<br /><br />Suggestions. The default sharpening in the eversmart/oxygen software is really really bad. Make sure they leave it off. Scanner noise can be smoothed out by scanning at twice the optical resolution of the scanner then downsampling to your desired output resolution. The eversmart pro has an optical res of 3175 SPI, but it's stepping motor is very precise so scanning at 6350 SPI will also likely get you a very small gain in resolution. Problem is that 4x5 at 6350 SPI is a rediculously big file, even in 8 BIT color.<br>

Hope I've helped.</p>