Film vs Digital - Color Rendition

<p>Rishi,</p>

<p>I don't understand your formula, or why you need it. The system resolution is lower than or equal to the lowest of the potential limiting factors. If the recording medium can register 100lpm but the lens 60lpm, system resolution is 60lpm. If the medium is still 100lpm but the lens is 150lpm then system is 100lpm.</p>

<p>I see your Velvia issue. I had many Velvia 20x30 Cibachromes done pre digital as wet prints. They truly don't compare well to my 135 format digital capture prints. I find it hard to believe such well respected pro labs were that far off. I have only ever seen scanning as another potential limiting factor, and a huge time factor.</p>

<p>"Most testers get 110-140"</p>

<p>Scott, do mean lines or line pairs per millimeter?</p>

<p>Rishi, I have a freezer heartily loaded with Techpan (120 and 135) and Technidol. To answer your question you can only buy it second hand today on ebay. If it interests you, you should stock up when it becomes available there.</p>

<p>Mauro,</p>

<p>The tests that I quickly looked at were lpmm, not lppmm. No doubling of the figures to wriggle out of this one. I actually ended up on the same site as Rishi did that first prompted his question.</p>

<p>How did you get your 350lpm film resolution figure? And why do you believe your lenses are able to resolve much higher figures than anybody else?</p>

<p>Everybody elses testing has come to the opinion that printers can print anything, just print bigger. Scanners can resolve film grain, there is a theoretical benefit to resolving further but it won't realise higher resolution figures. Film, well most of them, can out resolve any MF camera system. Lenses are the weak spot.</p>

<p>That pretty much shows your opinions to be 100% in contradiction to many testers, wonder why that is?</p>

 

<p>Ok. If it is line per mm then those test are vastly wrong. Here is an easy check: On the best axis my Coolscan 9000 resolves 160 lines per mm from my TMAX film, yet the film clearly resolves more than what the scanner can capture (at least 200 lines per milimiter).</p>

<p>For 350 lines per mm I shot a resolution target with Techpan and the RZ67, developed on Technidol and evaluated on a 400x microscope with diffuser.</p>

 

<p>Rishi,</p>

<p>Read the third post by 49 in <a href="http://www.apug.org/forums/archive/index.php/t-87068.html">this thread.</a></p>

<p>Scott, I truly hope you maintain a good discussion so I will keep responding.</p>

<p>First, resolving grain does not indicate resolution stops there. Not by a long margin. e.g. you can resolve TMAX 400 pushed grain at about 100-130 lines per mmm yet resolution clearly goes to 200 lines per mm.</p>

<p>You are incorrect with your assumptions about lenses. You don't have to take my tests for it. e.g. Phase One's IQ180 resolves 190+ lines per millimeter - that implies that the lenses resolve comfortably above 200 lines per millimeters; thus the lenses comfortably out-resolve all your current films.</p>

<p>Scott, I also remind you that I conducted my tests myself. Pointing to a link of a person that can't resolve 40 lpmm with TMAX can be avoided. You are welcome to borrow the film from me or Rishi.</p>

Awesome Scott, thanks for that link. Haven't had a chance yet to give it a read but will shortly. The formula I quoted is

in any basic optics book that covers resolving power.

 

As to your comment about 'seeing my Velvia issue', I'm confused. I was saying my 135 Velvia print is much more

pleasing than my 135 5D print. To be fair I'll try to print the Velvia at 24x36 shortly. But from the 100% crops I posted

above of the formats upscaled to 24x36 @360 ppi, the 5D is devoid of fine detail the Velvia retains.

 

If you're talking about a 5D MII, sure I'm willing to believe it may look nicer printed at 24x36 than 135 Velvia.

Remember when comparing my Velvia print to a wet print though that I applied lots of selective, masked input &

content sharpening, which I'd imagine would give it an advantage over traditional darkroom printing.

 

Also, good point about printing larger. And l agree it's well accepted that lenses are the limiting factor when it comes to film. Possibly also with the 5D MII.

 

Rishi

<p>Here you can see some of my test with the RZ67 using pushed TMAX 100, Provia, Velvia 50 and Velvia 100. This test shows the limitations of the Coolscan 9000 (not the film or the camera):</p>

<p><br />http://shutterclick.smugmug.com/Photography/Resolution-and-Diffraction/6302153_PLzKe#407096809_FeZJ4-O-LB</p>

<p> </p>

<p>"Also, good point about printing larger. And l agree it's well accepted that lenses are the limiting factor when it comes to film. Possibly also with the 5D MII."</p>

<p>Rishi, you are incorrect here. Simply look at the results of the 7D (pixel density of a 46MP 35mm camera) and you'd realize that 35mm lenses far outresolve the 5DII, the Coolscan, Imacons, Drumscans, and films.</p>

<p>Rishi, since you are scientifically disposed, this is the easiest way to test lenses without even developing film:</p>

<p>Rent an RZ67, set up your resolution chart, but instead of shooting film, install a 400x microscope on the fresnel (you can remove the glass). Enjoy.</p>

<blockquote>

<p>Rishi, you are incorrect here. Simply look at the results of the 7D (pixel density of a 46MP 35mm camera) and you'd realize that 35mm lenses far outresolve the 5DII, the Coolscan, Imacons, Drumscans, and films.</p>

</blockquote>

<p>All of my lenses, cameras and films out-resolve my photographic skill so who cares?!</p>

<p>They all certainly outresolve my printer and needs as well.</p>

<p>Mauro,</p>

<p><em>"Rishi, since you are scientifically disposed, this is the easiest way to test lenses without even developing film:</em></p>

 

<p><em>Rent an RZ67, set up your resolution chart, but instead of shooting film, install a 400x microscope on the fresnel (you can remove the glass). Enjoy."</em></p>

<p>Bob Atkins gives a perfect description of why testing like you suggest can be extremely misleading. <a href="http://www.bobatkins.com/photography/technical/resolution.html">Aerial image lens testing</a> figures are not accurate and give wildly optimistic resolution numbers.</p>

 

<p>He is right especially past the diffraction limit at f2.8 or f4. Still these come into effect at 400+ lpmm.</p>

<p>All these limits are way pass the limitation of films and sensors. Look at resolutions obtained digitally today: the P180 (190+ lpmm) or the 7D (230+ lpmm). Lenses are not the bottleneck - film and sensors are.</p>

<p>Film is the critical path: Ektar's limit is 150-160 lpmm, Velvia's limit is 160-180 lpmm, TMY2 is 200 lpmm, and TMAX is 200-220 lpmm.</p>

<p>And down the chain:<br>

Coolscan 9000's limit: 160 lpmm in one direction and 145-150 lpmm in the other.</p>

 

<p>Rishi - I absolutely love your cityscape, and I don't think I would care if I had to step back a few feet while viewing the print. That's what I call color rendition!</p>

<p>Rishi - I'm curious about what RAW converter you used for that shot. The "painted" look you got at that magnification reminds me of Canon's DPP, and is one of the key reasons I use ACR. ACR yields significantly higher resolution and better fine detail from the same RAW file in my experience.</p>

<p>Could be one of the differences between your experience and Scott's. (Of course there are also other very large differences. If I remember correctly he's using a 21 MP camera, and then there's distance to subject, lenses, the total lack of any possible air turbulence in Scott's shot, etc.)</p>

<p>One other comment on the lens resolution sub-debate: you guys are speaking as if resolution is a fixed number. It's not. It varies with detail contrast. If you have a high contrast target and are looking for any additional detail (i.e. MTF10) then lenses comfortably out resolve films and sensors. If you have a normal or low contrast target and are looking for detail that will make an impact in print (MTF50) then the lens in question may not out resolve the sensor or film.</p>

<p>Also, final image resolution is going to be less than the weakest component, not equal to. But at the same time if the strongest component is made stronger, final resolution will increase but will never reach the limit of the weakest component. That's a lot different from the way most people think about resolution and the imaging chain. (If I can dig up the links to the web pages which explain all of this in detail, with the relevant formulas, I will.)</p>

<p>For all of these reasons the difference between excellent, good, and mediocre lenses is noticeable even when using an old 6 MP sensor. The differences can be night and day with modern sensors, more so than with even the best films because digital sensors have a higher and flatter MTF curve than film. They pick up differences in MTF/contrast between lenses more than even Velvia 50.</p>

<p>DLT -- good to have you back on board. And, shucks, thanks for your compliments!</p>

<p>I exclusively use Adobe Lightroom & Photoshop, so, yeah I've been using ACR ever since they got their Canon-emulating DNG profiles (b/c previously Canon's color rendering was much better than ACR 3, but that's another story).</p>

<p>I typically begin by performing some content-aware input sharpening using limited masking (built into LR, of course), and then will selectively apply more sharpening using masks in PS if necessary. </p>

<p>This image was actually 6 different exposures (taken over the span of ~45 minutes... it was a ridiculous sunset, & that final image actually understates what I remember seeing) merged largely by hand-painted masks in PS; each of those was input sharpened, then the final image was content-sharpened, and then finally up-res'd & print-sharpened for whatever specific printer/size combo.</p>

<p>What you see there as the 'painted' look is Genuine Fractals upscaled & then print sharpened by LR for 360ppi. That's why it looks so, er, weird. Less visible in the actual print (print sharpened files look much worse on monitors than they do on the actual print).</p>

<p>The 5D did pretty well... 24x36 was cutting it; would definitely feel more comfortable selling it as a 20x30 print. Would feel even better upping the resolution of my shooting format :)</p>

<p>Responding to resolution stuff in a bit...<br>

-Rishi</p>

<blockquote>

<p><em>Bob Atkins gives a perfect description of why testing like you suggest can be extremely misleading. <a rel="nofollow" href="http://www.bobatkins.com/photography/technical/resolution.html" target="_blank">Aerial image lens testing</a> figures are not accurate and give wildly optimistic resolution numbers.</em></p>

</blockquote>

<p>I'm not saying Bob Atkins is wrong; however, he did pick an MTF level of 1%. At 10% or higher the conclusion would be exactly the opposite. I guess I'll just have to wrinkle my film plane to match the wavefronts, but that belongs to the flatness thread.</p>

<p>I think Mauro & Scott/I are disagreeing b/c of what you mention Daniel. Also, Mauro, I didn't say *I* personally believed that lenses were the limiting factor; I was simply repeating what I see rehashed over & over on the web. Personally, I haven't done the tests so I'm not qualified enough to know.<br /><br />Even if the lenses have higher resolving powers (read: higher response at any given spatial frequency) than the medium (e.g. film), that doesn't mean that it won't limit the highest perceived resolving power of that medium. If one believes the 1/Rsystem = 1/Rlens + 1/Rsensor equation, then a 320 l/mm resolving lens & a 160 l/mm film will only resolve ~107 lines/mm. That's all I was trying to say. Mauro, are you saying that your lenses resolve well in excess of 320 lines/mm? Do you agree with my theoretical calculation that your lens would've had to resolve 2400 lines/mm to get our figure of 150 lines/mm for Velvia 50?<br /><br />Is that equation right? DLT -- any thoughts?</p>

<p>Just to clarify: MTF10 is the spatial frequency where the medium shows 10% response, correct? Norman Koren's site says that the old definition of 'resolution' is the frequency where MTF is 10% or less (sometimes 2%-5%). So I guess all the variability we're seeing could just be based on different takes on 'resolution'. There's a lot of subjectivity here.</p>

<p>DLT, or anyone -- how do I measure the 'response' of that Imacon scan? I need an equation... (if you go back to page 30, the image is there along with some Photoshop measurements of the black & white lines at what I think is the limiting resolution of the film+lens)</p>

<p>I'm just getting some 1:2 & 1:20 contrast resolution charts printed to test along-side my 1:1000 chart. I'll be curious what the actual contrast of those charts are under indoor lighting... then will test some 35mm & MF formats shortly.</p>

<p>-Rishi</p>

<p>Even if lenses do 'outresolve' film or sensors b/c their MTF10 is a significantly higher spatial frequency than the medium, that still doesn't mean a lens with an even higher yet MTF10 won't yield better results. Because of the 1/R equation (if you believe it!). </p>

<p>Else, how do you explain this:<br>

<img src="http://rishisanyalphotography.com/ForumPostFiles/photo.net/24-70vs70-200_Sharpness@70.jpg" alt="" width="800" /><br>

<a href="http://rishisanyalphotography.com/ForumPostFiles/photo.net/24-70vs70-200_Sharpness@70.jpg">Full-Resolution Image</a></p>

<p>Or this:<br>

<img src="http://rishisanyalphotography.com/ForumPostFiles/photo.net/EF-S18-55vs17-40L.jpg" alt="" width="800" /><br>

<a href="http://rishisanyalphotography.com/ForumPostFiles/photo.net/EF-S18-55vs17-40L.jpg">Full-Resolution Image</a></p>

<p>View 'em at full res to appreciate the differences.</p>

<p>There's nothing really special here; the-digital-picture.com routinely does more objective tests like this & the fact that they see large differences between lenses means lenses are <strong>a</strong> limiting factor (even if they may technically 'outresolve' film), and along with the medium, contribute to a system resolution via some relationship like that 1/R (perhaps with an exponent now that I've read Bob Atkins' description) equation...</p>

<p>Mauro, are we in agreement? Just wanted to make sure we're understanding one another.</p>

<p>-Rishi</p>

<blockquote>

<p>Rishi: Is that equation right?</p>

</blockquote>

<p>Yes the equation is right. And to get 150 l/mm from 160 l/mm film you need a lens that can give you 2400 l/mm. Of course such a lens does not exist.</p>

<p>Google for "Fujifilm Dataguide" and download the pdf. On page 128 of the pdf document you can read how Fuji determines the resolving power of film, or the combination of film+lens.</p>