Why do negs always scan noisy?

[adrian_sancho]

<p>Does anyone know why negs look like such crap when scanned on a film scanner? I recently did some night shots on Fujicolor 100, and you'd swear I was using the old 1600 film. When I scan positive film, I don't get this insane grain. As a matter of fact, I've seen 400 ISO chromes look better than this. I'm scanning on a Minolta Dimage Scan 5400 scanner. Here's a 100% sample below. Any info on why this is happening will be appreciated, thanks.<br>

<img src="http://stashbox.org/1109218/scantest.jpg" alt="" width="600" height="800" /></p>

 

[yoshio]

<p>Hello<br>

part of it is grain and another part is the differences between the colour channels. I suggest you split red green and blue within your editor (the channel tool in photoshop), this exacerbates the issue.</p>

<p>This is perhaps a when you realise that you have enlarged this from 5400dpi scan and are examining it on screen at about 100dpi then you should expect to see quite an amount of grain. You could print that to 27 inches wide and still be printing at an optical 300dpi</p>

<p>That is about x18 magnification: that its that good is quiet acceptable if you ask me. What does a typical DSLR stood in the same place covering the same image in a single frame get?</p>

<p>Not as good I'll bet</p>

[dave_redmann]

<p>In many night shots, the subject-matter's dynamic range (that is, the range of brightest to darkest things in the frame) is much greater than you have in most day shots. So unless you expose so much that you totally blow out (make white) the better-lit areas, the darker areas will be very under-exposed. Under-exposed areas on negative film are going to look grainy.</p>

<p>Also, as already pointed out, taking a 5400 ppi scan and displaying part of it at 100% on a monitor of (in my case) 90 ppi or so is equivalent to a print of 57 x 85 inches (1.4 x 2.2 m). I don't care what film you use or how you expose, process, scan, and print it, if you print a 35mm frame that large, unless you're much farther away than I am from my monitor, it's going to look quite grainy.</p>

 

[coomber]

<p>Where is everyone getting the 5400ppi from, that number was the model of scanner used wasn't it? However, Adrian, Dave's comment above about dynamic range still stands very true, is this a problem in all the negs you've scanned or just in shots like the one above?<br /><br />I don't tend to use neg film much so I'm not sure if I can be of much more help!<br /><br />Craig. </p>

[Dave Luttmann]

<p>The Minolta 5400 scans at 5400ppi. There was also the Minolta 5400II, which also scans at 5400ppi. And I agree with the others....viewing a 5400 ppi scan at 100% on screen is like looking at a 5 or 6 foot wide print. No wonder we see grain. But yes, chromes will scan with less grain and noise (although you do end up with the Fuji pepper look) than negative film....but then you're also losing a good 5 stops of latitude by using chromes....even Astia.</p>

[bob_sunley]

<p>Some colour negative films have 6 layers of light sensitive emulsion, two for each colour. One of each pair is a very grainy high speed layer, which is why you get grainy under exposed negatives. Expose the other layer and the grain is dramatically reduced.</p>

[robert lee]

<p>It's a number of things. First, negative film is inherently noisier than slides. Second, Fujicolor 100 isn't a particularly low grained emulsion. Third and most importantly, negative performs poorly when it's not given sufficient light.</p>

<p>The following shows two different exposure settings on Superia 400 from the same camera and on the same scene.<br>

<img src="http://static.photo.net/attachments/bboard/00C/00CQUe-23923584.jpg" alt="" width="521" height="800" /></p>

<p>The problem with night scenes is the dynamic range is huge. Meter for the brighter parts of the image gives you a muddy noisy mess everywhere else. Meter for the shadows, however, and anything a bit brighter gets blown out (from reciprocity failure and no, compensation doesn't really work.) Pick your poison.</p>

<p>Oh, almost forgot. You can always exposure blend or use the more tradition HDR approaches. It's easier done with digital capture but film can work too.</p>

[sfcole]

<p>Adrian,<br>

Since night shots often don't have a "correct" exposure value in the way that a typical daylight shot would, is it possible you've underexposed for the look you want and brought up the levels in Photoshop? In that case, grain will become very noticeable, just as it will whenever one underexposes.</p>

[rishij]

<p>I'm going to put forward my hypothesis, & I'd love someone technically-minded to critique it.</p>

<p>It's well known that negatives record a wider range (9-14 stops, depending on who you ask) than slides; however, the negative then 'encodes' that range into a dynamic range of the medium smaller than that of slide. Why? </p>

<p>Because negatives at best can have a dynamic range on the medium of about 100, since their Dmax is at most ~2 (10^2). Velvia can have a Dmax of 4, but taking into account where information starts disappearing on both bright & dark ends, the real usable density range is ~3.7. So that's 10^3.7 or a luminance ratio of 5000 (Luminance Ratio = 10^D, where D is the density range of the medium). 5000:100 for slide:negative. The higher luminance ratio of the slide, given the same subject input range, allows more levels in the medium (film) to represent a given tonal input range. Let me quote Robert Feinman on this from <a rel="nofollow" href="http://www.tomshardware.com/forum/34492-5-scene-range-dynamic-range" target="_blank">here</a>:</p>

<blockquote>

<p><em>"Now once we have established the scene range we have the other factor which is how fine a gradation between values can be record. Film being an analog medium the answer should theoretically be infinite, but in </em><em>most cases it is usually considered to be some small finite density change like .02 or .03 on the film."</em></p>

</blockquote>

<p>Since a larger scene range is encoded into a smaller density range in negative film, the finest unite of finite density change discernible in the film will have to encode a larger range of scene luminosities in negatives as opposed to slides. So one might use the analogy: negatives have a lower effective 'bit depth' than slides. Let's run with this analogy & say negatives are an 8-bit capture medium & slides a 16-bit capture medium (because slides use more tones to represent the original dynamic range of the scene given its larger density range to which the original scene is mapped). Now, I quote Bob Atkins:</p>

 

<blockquote>

<p><em>"As an example you can take a 16bit image which represents a dynamic range of 4.0 and convert it to an 8 bit image which represents the same 4.0 dynamic range, but the 8 bit image will have only 255 levels and the 16 bit image will have 65536 levels. Since a DR of 4.0 is a ratio of 10000:1, each step in an 8 bit image will represent 39 "units" of linear brightness - or 0.015 density units, while each step in the 16 bit image represents 0.15 "units" of linear brightness or 0.00006 density units."</em></p>

</blockquote>

<p>So, though you have less recorded scene range in slides, you have finer/smoother tonal gradations since you have more signal (some arbitrary density range of dye clouds in the film) to represent a small, finite, scene range. Given the inherent noise of grain/dye clouds, I kind of see this as great signal:noise in the slide vs. the negative. And that would explain why when you attempt to re-introduce contrast into the scanned negative in post-processing software, you start seeing more & more of the grain (noise). Because the signal:noise was low to begin with. This is also why defects in the film base (noise) become more obvious in post-processed negatives than their slide counterparts; because compared to the lower signal in the negative, the defect = formidable noise. Upon amplification of the scanned signal in post-processing with contrast adjustment, that noise is also enhanced. Of course this problem is largely solved by infrared channel subtraction in scanning.</p>

<p>Does this make sense?</p>

<p>One question I've always had: during the time of exposure, do slides & negatives capture the same amount of scene range in the latent image? Either way, after development chemistry, I believe one can argue that, since a larger scene range is represented by a smaller density range in negative film:</p>

 

<ul>

<li>An increase of 1 stop in the scene recorded by a negative is represented by n * d, where d is the density of the negative for the darker scene element, & n is some arbitrary value > 1.</li>

<li>A decrease of 1 stop in the scene recorded by a slide is represented by i * [n * d], where i > 1, where d is the density of the slide for the lighter scene element.</li>

</ul>

<p><br />Therefore, a greater signal change occurs in the slide as compared to the negative for the same change in scene brightness. Therefore, overall, slides should inherently have a larger signal:noise ratio.</p>

<p>Which would explain why I also always see what the OP sees: scanned negatives are always noisier than slides. Even Ektar 100 exposed at EI 50 compared to Velvia 100.</p>

<p>-Rishi</p>

<p>P.S. I'm fairly sure I'm not the 1st one to put forward this hypothesis; I just can't find any references on the internet. If you know of any, please share.</p>

[rishij]

<p>BTW, great comparison Robert. That's exactly why I dial in an ISO 1 stop slower for my negative films. And even then tend to 'expose to the right' (in digital terms, hehe). Meaning expose for the shadows.</p>

<p>-Rishi</p>

[rishij]

<p>So this might also be because the dynamic range of the output voltages from the CCD when negatives are scanned (with their limited densities) are not well matched to the A/D converter, since that A/D converter must also handle the larger range of voltages output by the CCD for slides.</p>

<p>But still, I think it has to do with the limited dynamic range (read: density range) of negatives vs. slides (again I clarify: I'm not talking about the scene range recorded, which is larger for negatives!)</p>

<p>Why? Because barring infinite tonality within the dynamic range of a medium, the smaller the dynamic range of that medium, the more limited its capability to accurately represent the original signal.</p>

<p>For example, say you have a scene with a luminance ratio of 8:1. You have 8 objects in it each with relative luminance of 1, 2, 3, 4, 5, 6, 7, 8. You now have a CCD that only outputs voltages in 1v increments (of course that's silly, it'd have much finer increments, but this is for the sake of making the argument easier to understand. At some level, the voltage output <strong>is</strong> discrete, especially in light of whatever is next in the electronics that reads the CCD signal; i.e., the A/D converter has a limited sensitivity to changes in input voltage beyond which differences in voltage cannot be resolved)... this discrete nature is imposed to mimic the limited discernible density change in film. You now have a CCD that's voltage output range is 0-4v, and another one's who's output range is 0-8v. The two CCDs represent the luminosities of said objects with the following output voltages (input luminosity-->output voltage):</p>

<ul>

<li><strong>4v CCD</strong>: 1-->1, 2-->1, 3-->2, 4-->2, 5-->3, 6-->3, 7-->4, 8-->4</li>

<li><strong>8v CCD</strong>: 1-->1, 2-->2, 3-->3, 4-->4, 5-->5, 6-->6, 7-->7, 8-->8</li>

</ul>

<p>Now, let's multiply the output of the 4v CCD by 2 to make it match the output of the 8v CCD (analogous to doubling exposure in Photoshop if this were digitized data). We now get the 8 luminosities of the scene represented, by the 4v & 8v CCDs, respectively, as:</p>

<ul>

<li>2,2,4,4,6,6,8,8</li>

<li>1,2,3,4,5,6,7,8</li>

</ul>

<p>Therefore, for a recording medium (CCD or film) with finite gradations in tonality, the medium with the higher dynamic range will represent the original signal better. I.e. slides represent the signal better than negatives for a given input scene range that fits within the usable dynamic range of both mediums.</p>

<p>In the end, I think it just boils down to the fact that slides use a larger # of dye clouds to represent a finite tonal change in the original scene than negatives do.</p>

<p>-Rishi</p>

[mark_denton1]

<p>Richie<br /> I'm definitely *not* as technically minded as you and struggled to follow all of the above so pls bear with me :)</p>

<p>I understand the fact that slide has a wider recording range than neg, but I'm still not sure how slide film's wider recording range than neg addresses the grain issue. Aren't these 2 separate things, ie, to use your digital analogy, one's about 'bit depth' and the other is about 'noise'?</p>

<p>I also wonder if the digital analogy is itself leading us astray. After all, colour neg is an analogue medium; it still fits the 'original signal' of a scene into a *continuous* range of colours/tones, albeit one that, is indeed a narrower range than slide film. And provided that the medium that will receive the projected image (ie the sheet of colour paper or the scanner CCD) from the neg is sufficiently sensitive to these colours then the fact that one recording medium records this information over a wider spectrum surely is moot?</p>

<p>In practice, i've certainly never suffered from banding problems or similar colour-sensitivity issues when using neg film either in the darkroom or when scanned.</p>

<p> </p>

[randall_pukalo]

<p>Look, everyone is missing the point - Negative film scans much worse than slide film. It scans like crap, to be honest, unless you have a dedicated minilab film scanner. No, a Coolscan 9000 wont solve this problem (I have one). Lets admit it, be honest, and get rid of the old myth that slide film is more difficult to scan than negative film. This crazy idea drives many new film users away from slide film without ever trying it, then they are let down when they see how badly negative film scans, then they leave their short stay in film.<br /> Not only is slide easier to scan, it scans VASTLY BETTER than negative film. The colors are vibrant, the colors are correct, and the image is very sharp with very little grain. It can be sharpened, much like digital, without the need to first run a blurring grain reduction filter over the image. And as long as you use a camera with a decent meter - I'm talking the equivalent of an $80 Olympus Stylus Epics meter, your exposures will be fine. And your scans - even with a cheap flatbed scanner - FAR better than what you will get with negative film.</p>