Film vs Digital - Color Rendition

[mauro_franic]

<p>At any point before the competition is closed, any participant may re-submit their entry any number of times (only the last one will be considered). Hopefully this will promote exchange of ideas.</p>

[rishij]

<p>Mauro, this might be a fun game, but you could also go about this rather objectively. Perhaps that wasn't your purpose for this thread?</p>

<p>For example, you could shoot an IT8 target with 288 or so color patches on a dSLR vs different films, take the spectrophotometric measurements of that chart evenly illuminated with a standard illuminant (D50?), develop the RAW file/scan the films using pretty consistent technique (e.g. no auto corrections, etc.), then use Argyll or Lprof tools to build 'scanner' profiles (that include tone response curve corrections) for your RAW digital capture & all film captures. Then apply these profiles to each image. Theoretically, they'd all look similar at this point, & you'd have an objective basis for comparison of a lot of colors.</p>

<p>To make a game out of it, you could add grain using RealGrain or something to the digital capture, then make people guess which is what. Or remove grain from the film scans, etc.</p>

<p>Anyone think this might be worthwhile? I've considered trying it; just haven't gotten around to it.</p>

<p>Cheers,<br />Rishi </p>

[mauro_franic]

<p>That could have been one approach but I decided for a lively more day-to-day target that may spice people's interest (at the expense of less exactly quantitative approach).</p>

<p>I also don't want to give too much away. The flowers and the face are plenty to experiment with the concept and value of color-in-a-can.</p>

<p>I hope to see your rendition soon.</p>

[mauro_franic]

<p>After the contest is over I will still keep sharing all the information I can and they will be plenty for discussion.</p>

[eddy_d]

<p><em>Sorry, but digital capture is technically capable of far smoother tonal transitions than film could ever hope to achieve, and the very fact that we can use it to imitate the limited range of colours that a particular film reproduces should be ample proof of that.</em></p>

<p>If you are shooting digital then why would you want to imitate film? If you are shooting digital then why do you try to get your digital photos to look like film photos? if you are shooting digital they should look like digital. If digital is great the photos should be perfect already with no need for post.<em><br /></em></p>

[rishij]

<p>Rodeo Joe: interesting points. Your analysis of the # of colors that can be represented on film by an area the size of a pixel of course assumes that your film is only using that small area to represent a color. It may be using more (which, of course, lowers its effective resolution). So I really wonder how objectively this can be examined.</p>

<p>A 24MP (pretty much the limiting resolution of film w/ very high contrast subjects) scan of 35mm Velvia translates to a 6 micron x 6 micron area of film. That's 36 square microns. How many dye clouds can fit into that small an area? And wouldn't you multiply that by 3 for the three emulsion layers?</p>

<p>Also, you say: </p>

<blockquote>

<p>The colour in film is captured by 3 and only 3 discrete colours.</p>

</blockquote>

<p>Yes, but the absorption bands of these dyes are incredibly broad. Like the human eye. So technically, each dye is responding to a large number of wavelengths. The transmission profiles of the R, G, & B filters on Bayer sensors are really strange. They have sharp peaks at very selective wavelengths. Colors are then interpolated. I wish I had a link to a transmission profile but I can't find one right now. Film can also be seen as an interpolation process. Which one is better is up for debate but technically having broader absorption profiles necessarily means you're recording more information. Again, whether that's useful information or not to the human eye in the final product? I don't know.</p>

<p>Also, you say:</p>

<blockquote>

<p>In addition, each film dye cloud has a fairly uniform density, so shades of colour have to be represented by additions of densities or by the natural dithering of randomly scattered dye clouds. </p>

</blockquote>

<p>Actually, I think you want very uniform properties to the 'primaries' when you're mixing colors in either an additive or subtractive process. Purer CMY pigments lead to a larger gamut of colors achievable using a subtractive method, no? And purer RGB primaries are used in wide gamut monitors, no? As I understand it, the purer the primaries, the more accurately you can mix them to get a wider gamut of colors than may be achievable with primaries that are 'polluted' by wavelengths of another primary. So basically what I'm saying is you want the dyes to have uniform properties so you get an accurate representation of the original color information when you 'add' them.</p>

<p>I may be mistaken about that whole purer primaries & wider gamut thing. Please correct me if I'm wrong.</p>

<p>Cheers,<br>

Rishi</p>

<p> </p>

[scott_ferris]

<p>Rishi,</p>

<p>I think that is a far more worthwhile approach and it should silence any film "feel" comments to boot, please do it!</p>

<p>Eddy,</p>

<p>Three times I have had paying clients ask for "film looks". Once was for Kodachrome, but what they actually asked for was "that seventies style with rounded corners and funny colours", when I showed them examples they were thinking of Kodachrome. Once was a very opinionated regional magazine editor who wanted Velvia 50 style high contrast and very saturated colours. The last was not for a specific emulsion but a grainy B&W set of images. All three got their images as they wanted them, all three were shot native digital. Were they close enough to the exact emulsions to satisfy some of the posters here? I doubt it, but they were close enough to get me paid by happy clients, one of which is an imaging professional.</p>

[huang_shao_hui]

<p>Scott Ferris says<br>

<em>At the moment yes, but I know how you and Les continue these threads, if somebody gets close then you will look deeper, if they get closer again after actually showing them what they are trying to copy then you will look ever closer, you are very happy to compare pixel level posts, well you have been before.</em><br>

<em>Just as an example, we know Dave can't be considered impartial and how in the hell could he enter a competition that he is judging? Even if you say he won't judge his own work that means he is not judged as the others were! Amazing........</em></p>

<p>I always feel amazing how people like Scott Ferris have such a deep passion in this sort of threads which they themselves criticized till the end of the day. If when challenge for facts, the standard tag line would be</p>

<p><em>For my work my 135 format digital camera vastly outperforms my film use up to and including 6x7, I still drag the 6x9 out every now and again but more for the fun of it than the outright detail or colour or "feel".</em></p>

<p>Yeah, very true, because the above statement your '<strong>digital camera</strong> vastly outperform <strong>YOUR film</strong>' And from what I know you do not represent anyone but yourself. Remember always, your experience is, well your experience ONLY... I don't think Mauro forces you to accept his experience as <strong>THE ultimate </strong>experience for mankind? :)<br>

To Scott et al. If you're not interested and if you feel so much unfairness and outright silly or whatever, then just ignore it. Let those who want to have the fun participate rather than fill up this thread with useless noise. Why not start <strong>your OWN TEST</strong>, comparison blah blah blah at your DIGITAL forum, and whack film however you like. Show to the world how <strong>impartial </strong>you are by hiring<strong> MR from LL</strong> (why am I even talking about this joke material?) and his gang be the judges. Won't that make your day? Why bother to create endless amount of noise in here? Last time I check, the competition was set up Mauro, so if you think it's not fair, don't enter. In fact, don't even bother since he's a 'track record' with his accomplice to make digital look bad. </p>

<p>Go get a life.</p>

[rishij]

<p>I really think everyone here should be more open to criticism & challenge, & not respond angrily when one person criticizes another as long as the criticism is objective.</p>

<p>Criticism & being challenged is how you grow. </p>

<p>Since Mauro, whose work I often admire, titled this thread "Film vs Digital - Color Rendition", it's reasonable for someone casually strolling by to assume that this might be an objective test of the color rendition of the two formats. When one then reads the thread & feels it's not objective at all, he/she might be inclined to point this out. At that point Mauro may say 'well, your point is well taken & I meant this as more of a subjective test' or he could defend his feeling of it being objective. </p>

<p>For example, I pointed out what may have been a more objective test. Mauro said he'd considered it, but decided to go with this fun approach instead. End of story. No harsh feelings.</p>

<p>You all need to just chill out.</p>

<p>Personally I feel that more objective comparisons of film vs. digital <strong>are</strong> possible. In fact, Mauro's done a number of them in the past. So instead of the back & forth yelling, those with opinions would do better to simply express them in a calm objective manner and then, if they choose to make a bold claim re: one format over the other, perform their own tests and show us the results. We can then, as a community, engage in banter over whether or not those tests are valid. That's pretty much how good science is done. I do intend to do the experiment I outlined above, just to see how far each format can be pushed. In the meantime, I'll enjoy this thread for what it is: and that is certainly <em>not</em> an objective comparison of the color capabilities of the different formats. </p>

<p>Rishi</p>

[tony_leinster]

<p>Rishi<br>

one final attempt! As you say the problem comes when a thread is labelled "Film vs Digital" or "film vs 5DII" or similar. Then when anyone questions the criteria the answer is "well it's not really film vs digital it's just....". Why use that sort of title then? If Mauro wants to just have some fun looking at trying to reproduce a film "look" then just say so. If you use a confrontational title then you will get a confrontational response, especially if the methodology is flawed. As I have said many times, I have no particular axe to grind either way, I just like to see fair play.</p>

[radek_pohl]

<p>Daniel Lee Taylor:<br>

<em>"Note that the 7D has finer resolution of relief texture details, which are color, than 35mm velvia 50 on an Imacon scanner"</em><br>

And where are blue river lines gone ? - river network drawing is clearly visible on velvia scan, but missing on 7D image (actually not missing, but drawn with background color tone instead of blue)</p>

<p><em> </em><br>

<em> </em><br>

<em> </em></p>

[rishij]

<p>Tony, if it were me, I would've chosen a different title for the thread :)</p>

<p>I also have no axe to grind b/c I use both formats & have researched scanning/processing of film (hence I use a 'digital workflow' for my film) enough so that when I ask people to look at a bunch of prints from my Epson R2400 & tell me which one is film or digital, often they have a difficult time (except for dead giveaways like reciprocity failure or pepper grain). </p>

<p>I do like objective tests though b/c they help me push these technologies to their limits. Something I derive pleasure from :)</p>

<p>That being said, this thread here is about as objective as me asking you to go <a href="http://staff.washington.edu/rjsanyal/photoblog/2010/03/09/trabant-showing/">here</a> & tell me which ones are film & which ones are digital. If done properly, it shouldn't be a dead giveaway until you start pushing the limits of the format (100% crop or large prints), at which point you will see grain for film and jagged edges & sharpening halos for digital. Or something like that; I don't know if I explained that well.</p>

<p>This thread doesn't push the limits of either format as far as I can tell (perhaps Mauro shall prove me wrong). The more objective test I outlined earlier might; or it might show that both formats are equally capable -- I won't know until I do the test!</p>

<p>-Rishi</p>

[StuartMoxham]

<p>I wonder what the conclusions will be from this thread.</p>

<p>Without the film scans or reference images it is near impossible to match them with digital as you have no reference. It would also be impossible to make Porta 160 look like Velvia without Velvia reference images and equally difficult to make make Potra match a DSLR image.</p>

<p>For me I feel people should shoot whatever floats their boat and stop worrying film VS digital. Some prefer one medium over the other that does not make them wrong either. Some like to take advantage of both mediums and enjoy shooting both.</p>

[Karim Ghantous]

<p>Certainly an important topic, but it would have been more convenient to just do another comparison rather than have a competition. The comparison would then spur on others to try the exercise for themselves. Just IMHO.</p>

<p>I agree with the question of whether or not film is worth the trouble. The problem is that there is no answer to agree with. It's all subjective and chaotic. Which trade-offs are easier to live with?</p>

<p>BTW do web browsers support colour profiles yet?</p>

[rishij]

<blockquote>

<p>BTW do web browsers support colour profiles yet?</p>

</blockquote>

<p>If you're on a Mac, then yes for Chrome, Firefox & Safari.</p>

<p>If you're on a PC, then yes if you're using Apple Safari. And possibly if you use Firefox's plugin, though I'm not sure if that gets you active translation to the monitor profile. Usually applications in Windows take the lazy route & convert to sRGB & just ignore your monitor profile.</p>

<p>And, yes, that's a pretty sad state of affairs for PCs.</p>

<p> </p>

[Dave Luttmann]

<p>Rodeo, dye layers have nothing to do with tonal accuracy, or bit depth....and nothing to do with the OP's post.</p>

[rishij]

<p>Certainly nothing to do with bit depth, but I've always been curious about how large an area (e.g. in square microns) is needed to represent thousands of tones? We started broaching that topic in <a href="00ROOo">this</a> thread, but never really hashed it out.</p>

<p>The answer to this is relevant to film's native 'resolution' as one might compare it to a digital camera's megapixel resolution. (i.e. you take the square root of this area & then divide 36,000 microns by this number, in microns of course, to get the effective 'pixel equivalent' count along the long axis; then do the same for the short axis, etc.)</p>

<p>Ok maybe not relevant to this thread, but, still very interesting!<br>

Rishi</p>

[maderik]

<p>Radek,</p>

<blockquote>

<p>And where are blue river lines gone ? - river network drawing is clearly visible on velvia scan, but missing on 7D image (actually not missing, but drawn with background color tone instead of blue)</p>

</blockquote>

<p>Lost in the algorithms of ACR which make the (reasonable) assumption that the human eye is much better at distinguishing contrast than color for fine details. Then again, that's like asking where is the green in the island of Sao Tome and Principe in any of the Velvia versions including Mauro's sharpened ones? Where are the ocean/land border lines near Port Gentil? The answer here is that they are lost in the dye clouds and scanner optics (and for Mauro's later version, the oversharpening.)</p>

<p>You can pixel peep these images all day long and find minor differences where A is better than B for <em>some</em> definition of better. And if you change the workflow (raw processor + sharpening or scanner + processing) and repeat, you will come out with slightly different pros/cons. So what's the point? Both can be very good but neither are perfect and it's then down to a matter of preference (e.g., which artifacts you can accept or not) and achievability (e.g., do you have the expertise in optimizing one or the other.) Hence the neverending debate.</p>

[charlie_novice]

<p>Mauro, it seems that the JPG from a CR2 raw file opened in DPP is not a neutral starting point, no matter how few adjustments you make.<br /> <br />I opened a CR2 file in DPP, in Picture Window Pro, and Raw Therapee. You can try RT; it's free. PWP and RT do not alter the exposure; the user decides what to do. DPP applies exposure and other changes when it opens a CR2 file. Perhaps I missed something in DPP.<br /><br />I discovered the difference when I opened your large JPG in Picture Window Pro. I cropped to the floral shot and applied a standard Velvia saturation action that a PWP user has contributed to the community. The result had the usual overdone green and such of Velvia, but it was not as different from the starting JPG as I expected.<br /><br />I suspect that even when DPP opens a CR2, it starts by applying all the choices you had in your camera's JPG menus. Anyway, there really is no "unbiased" digital image. The raw file is not a finished product, and canned adjustments or conscious post-processing happens. I suppose there is similar variation in developing film.</p>

[StuartMoxham]

<p>Well here is my guess for Tmax 400.</p><div></div>

[mauro_franic]

I will be traveling for one day and I may not be able to respond to questions until tomorrow.

[mauro_franic]

Stuart, please post the rest of the films. Great prizes await.

[mauro_franic]

Charlie, when you open the raw in dpp just set all the sliders to zero. Ignore any settings that dpp is importing from

your picture styles set in the camera.

[charlie_novice]

<p>"Charlie, when you open the raw in dpp just set all the sliders to zero. Ignore any settings that dpp is importing from your picture styles set in the camera."<br>

Mauro, DPP makes exposure adjustment(s) even when the sliders are at zero. Open a raw file in Raw Therapee or Picture Window Pro and you will see what an unadjusted raw file looks like. (Sorry, I do not have access to ACR).</p>

<p> </p>

[rishij]

<p>or do a conversion in the platform independent <a href="http://www.cybercom.net/~dcoffin/dcraw/">dcraw</a>... of course it'll have no idea what to do with colors until you feed it a camera profile.</p>