Jump to content

Memories of detail


Recommended Posts

Mauro, I never replaced my MF camera with a digital camera. I used them side by side. Sometimes MF was better.

Sometimes digital was better. Performance depends on lighting and other conditions since the two technologies render

images differently.

 

When I replaced my MF film camera it was with the 4x5 film camera that I still shoot regularly. No detail has ever been

lost. Nor have I ever lost a shot that film would have missed but digital could have captured easily, as is sometimes the

case.

 

I'll post my revised example this evening.

Link to comment
Share on other sites

  • Replies 102
  • Created
  • Last Reply

Top Posters In This Topic

<p><em>Scott, B&W conversions are difficult from digital because the data lost from the Bayer and AA filters cannot be recovered (that is in addition to the sensor's limitations).</em></p>

<p>The commonly held belief that Bayer filters cause a loss of detail relative to color film has been obliterated by observation. If Bayer did this then film would have higher low contrast and color detail than digital. This is clearly not the case when comparing within the same format. At 12 MP DSLRs start to resolve more low contrast and color detail (see the D2x sample at Les Sarile's site). At 18 MP and greater DSLRs exhibit considerably more low contrast and color detail, the very detail Bayer should destroy in theory. DSLRs resolve very close to their Nyquist limit pretty much regardless of brand or model. A careful comparison against Foveon sensors shows a small loss of detail, but clearly Bayer is much more efficient and effective that many people estimate from theory alone.</p>

<p>Relative to B&W film Bayer is not throwing away information, it is preserving information that B&W film by nature throws away. Try applying a yellow or deep red filter to a B&W film image after the shot.</p>

<p>AA filters cause a loss of sharpness more than anything else, which can be restored in post. I'm unconvinced there's any real loss of detail having looked at images produced by a DSLR with its AA filter removed, but I can't confirm or deny that without having one to test against a resolution target.</p>

<p>No, the reason why B&W conversions from digital can be challenging has nothing to do with resolution, Bayer, or AA filters, and everything to do with tonality. Years were spent perfecting the tonal curve of B&W film, and years of study and information are available on development which photographers can use to customize this curve in the lab. A straight conversion of a digital color image is typically flat and muddy by comparison. It takes some work, and some experience with classic B&W, to mimic the tonal response of classic B&W films.</p>

<p>A lot of the digital B&W work I see does not have that classic look regardless of how it stands on its own. (It may be excellent work, but the tonality still betrays it as different.) It is certainly possible to mimic that tonal response however. The best advice I can give on this point is to spend some time shooting both digital and B&W side by side, scene by scene. Every where you go take two SLRs, and everything you shoot, shoot both. As you develop and print the B&W (ideally yourself in a darkroom) develop and print the digital files to match the B&W. You'll get a solid feel for the differences between them, and how to get the digital to mimic the B&W tonality. Also, I think Silver Efex Pro is easier to use in this respect than the tools built into PS.</p>

<p>The upside with digital is that once you have the eye and technique needed to mimic the look, you can consistently get that look from any scene. My experience with B&W film is that it either looks great or looks terrible depending on how well the film and development is matched to the scene. Digital RAW gives you a pretty neutral capture of a scene regardless of the tonality in the scene. But then you have a host of tools in the RAW converter and in PS (levels, Shadows & Highlights, LCE, curves, etc.) with which to reshape the image curve to match the scene. You also can apply color filtration after the shot, a notable advantage in my book. You don't have to take notes and manage this before seeing the image on film, something Zone system adherents are religious about but which isn't always practical with roll film. You can just do it with image in hand.</p>

Link to comment
Share on other sites

<p>Daniel,</p>

<p>I agree. On average, I see more terrible B&W film work than good. Unfortunately, many people don't expose correctly for the shadows and compensate for highlights to suit scans. Thus, what we end up with is a combination of poor exposure, poor processing, and then poor scanning getting the worst of the worst.....blocked up shadows, clipped highlights and acentuated grain.</p>

<p>I used a mountain of TP when I used to hypersensitize it for astronomical photography. I started using it for some pictoral work but never really enjoyed the tonality of it. I prefered the look of Ilford FP4 and HP5.....TP looked stark in Technidol. I love the lack of grain in 6x7, but prefered the look of HP5.</p>

<p>Which brings me to my last point. I thought, back then, that I would like to have as little grain as possible in print. What I found was that with a 6x7 neg, even with FP4, there was minimal grain in a 24x30. In fact, when I look at scans from those old negs, I actually prefer them with that little bit of grain....it adds a bit of texture to the final image, giving the appearance of even greater detail. Therin lay my issue....while TP had more real detail in print, the FP4 looked like it had as much because of this fine texture.</p>

<p>Now, most of us add a bit of noise to images in prep for print. Adding this noise to a 5D2 image during B&W conversion in Silver Efex showed me that there was little more in the scans from my RB67 than the 5D2....at 16x20....and sometimes even 20x30. A bit of noise in the B&W really makes it appear more detailed. </p>

<p>I used a Pentax k20D a lot in Death Valley last year. I've made some 24x36 prints on HM Photorag 308 that were processed in Silver Efex. These looked superb.....despite the poor showing at 100% on screen.</p>

<p>While I love film for a lot of work, sometimes it doesn't matter. For this shot of clay in the dunes, at 24x36, I can see grains of sand with no problem. Processed in Silver Efex with the HP5 look if I recall. I reduced the grain somewhat to match what a 6x7 HP5 scan would look like.</p>

<p><img src="http://2.bp.blogspot.com/_mX5HKyiQAwQ/TDNDG1a9SGI/AAAAAAAAAGU/tCugD4nz784/s400/Deat+Valley+Dunes.jpg" alt="" /></p>

Link to comment
Share on other sites

<p>Mauro,</p>

<p>Given that your crop represents an area of film equal to 15.7mm sq and mine is from a sensor area of 3.45mm sq I think it is pretty amazing that the digital is anywhere near the MF film.</p>

<p>So for your consideration, here is a 15.7mm sq crop from the digital sensor. This now levels the playing field, we are comparing like for like. High end FF digital SLR sensor at 100iso and Tech Pan at 25iso.</p>

<p>I have done no processing to this crop, it is downsized slightly and converted in Save to Web and Devices but all with auto settings. No sharpening.</p><div>00XKuL-283043584.jpg.658411c69dd32382dc3ed3b2a4abbe65.jpg</div>

Link to comment
Share on other sites

<p>Daniel, current B&W films are panchromatic -sensitive to all colors of light. Whereas each pixel of a digital camera is sensitive to a single color. </p>

<p>If you use a color filter with a film camera then yes, that difference is gone.</p>

<p>It is true that the response of B&W films to different wavelengths has been tailored over years and years to excel in their application. I agree this is a big component of why it is difficult to replicate with a digital conversion.</p>

Link to comment
Share on other sites

<p>Marketing is very powerful. Look at the release of today's 46MP foveon camera:<br>

http://www.dpreview.com/news/1009/10092129sigmasd1.asp</p>

<p>Now under the same marketing technique, you would say the Coolscan 9000 is not 4,000dpi but 12,000 dpi. Would you?</p>

<p>Now from your 7D, you take a red pixel, a green pixel and a blue pixel... How do you produce three full RGB pixels? You can't.</p>

Link to comment
Share on other sites

<p>Mauro,</p>

<p>I'll post anything you would like. But what is it about the latest one you are not comfortable with? It is a crop from the same area as the film, both have been resampled to fit the 700px wide format, I don't think either is hurt more than the other that way. Either the scan, or the dslr image need to be resampled to match each other as they have different dpi/ppi figures.</p>

<p>Funnily enough I just saw that announcement <a href="http://theonlinephotographer.typepad.com/the_online_photographer/blog_index.html">here</a>. For me, none of this is about how I get there, it is what I end up with. When digital matched my film quality and smashed the cost per image equation there was no justification for shooting film. I don't care about the number of pixels or the grain and developer combination's, I care about the images. I wet developed B&W for years, I can replicate, to my satisfaction, any effect/mood/look on a computer far easier and cheaper (and environmentally friendlier!) than I could in the darkroom. I loved my time in the darkroom, but technology has moved on. My very low volume MF film images do still realise higher ultimate IQ than my "FF" digital, if you do all the things necessary to realise that potential. But the occasions when I need that additional IQ are very rare.</p>

<p>Interesting thread, thanks all, Scott.</p>

Link to comment
Share on other sites

<p>Scott, that is fine. Coincidentally my scanner produces 21MP from a 35mm frame. One-to-one relationship to a 20MP camera. Neither one has to be resampled.</p>

<p>It is interesting since people always want to see organic comparisons (instead of maps or resolution charts).</p>

<p>Daniel, did you eve develop the Velvia from the trip you took?</p>

<p> </p>

Link to comment
Share on other sites

<p>Mauro,</p>

<p>Here is a square crop at 1:1 for pixels. It is 700x700 pixels so from my camera (a Canon 1Ds MkIII) it represents 20.25mm sq of sensor area.</p>

<p>As normal I did nothing to it, opened in PS4 as 16bit ProPhoto spaced image, cropped to 700 pixel square and saved for web and devices at 75% quality. So it ends up being a 180kb, 8bit, sRGB, jpeg.</p>

<p>EXIF data is 1/200 sec at f4, 100iso, manual flash at 1/4 power. Lens was the much maligned (though I really like it) 50mm f1.4.</p>

<p>Hope this helps future threads too. Scott.</p><div>00XKxl-283085584.jpg.130b533bef5c081e7e1d586328404fa9.jpg</div>

Link to comment
Share on other sites

<p>Here are a few more treatments of my original eye photo, this time processed in Photoshop.</p>

<p>First, let's get some math out of the way. The size of the eye in the frame is significant. If the eyes in my frame are a larger part of the entire photo, for example, then my photo might appear to have more detail. I don't have Mauro's scan, so I can't compare the size of the eyes in his photo versus my own, but I'll offer size information for my crop and the full frame. Perhaps Mauro can do the same so we can get some idea as to whether the eyes are larger in one photo or another.</p>

<p>Sorry, Mauro, but I don't believe your claim that your crop is .04 percent of your entire photo (it's probably more like 4 percent), but you can verify this by presenting data and calculations similar to the ones below.</p>

<p>The whole frame is 5,616 by 3,744 pixels.<br /> The crop below is 584 by 419 pixels.</p>

<p>584 times 419 = 244,696 pixels<br /> 5616 times 3744 = 21,026,304 pixels<br /> 244,696 / 21,026,304 = 0.0116</p>

<p>Therefore, the crop is 1.16 percent of the entire frame. Conversely, the entire frame is made up of 86 segments each the size of this small crop. That's quite a bit of detail, IMHO.</p>

<p>Here is the original 584 x 419 pixel crop processed in Photoshop with no sharpening. I can see clearly pores in the skin, individual specs of eye makeup, and very fine strands of hair throughout the image.</p>

<p> </p><div>00XKzL-283099584.jpg.64aa02dcfc326c945d7d6b3876f47bb2.jpg</div>

Link to comment
Share on other sites

<p>Next, I used Photoshop's Bicubic function to upsize the crop to 700 pixels on the long side (the largest size that I can upload here). This is the unsharpened version.</p>

<p>Note that it does not have the freakish distortion of Mauro's attempt to upsize the JPEG file that I posted on Sunday. Everything is very smooth.</p>

<div>00XKzU-283101584.jpg.c2c08f685aede140a86c82177043fd0e.jpg</div>

Link to comment
Share on other sites

<p>Here is the upsized crop (700 pixels) with a modest amount of the Smart Sharpen filter applied. Lots of detail is visible. It is at least comparable and possibly slightly superior to the film scan. Again, we don't know the pixel dimensions of the eyes in Mauro's film scan, so a size differential could explain the increased detail if it indeed works out that way. If it turns out the my photo's eyes were SMALLER than Mauro's, then the detail in the digital capture is even more impressive.</p>

<p>Keep in mind that these shots were handheld on a sidewalk in mid-afternoon on a clear day. No tripod, flash, reflectors, or artificial lighting of any kind was used. The exposure was 1/250 @ f/5.6, ISO 200. This shot would not have been possible with a 6x7 camera loaded with ISO 25 file.</p><div>00XKzW-283101684.jpg.2be9231ceccdeab728c57b04db2e20ca.jpg</div>

Link to comment
Share on other sites

<p>Working on Mauro's first, multiple image, by my calculations, his tightest eye crop is very close to 1% of the full image area, +/- 5%. More than 20x the 0.04 figure he thought the crop represented. I got my first figures wrong by a similar magnitude though!</p>

<p>I did that calculation by cropping the full frame out of the montage, for an image 686 X 550 pixels, then the eye crop out of that is approximately 69 X 55 pixels.<br /> The full image is 377,300 pixels<br /> The eye crop is 3,795 pixels</p>

<p>377,300 / 3,795 = 99.4, or fractionally over 1%.</p>

<p>I only point this out in the hope that we can really put some inaccuracies and fallacies to bed in this thread.</p>

Link to comment
Share on other sites

<p>By using a simple ruler to measure Mauro's crops on the screen, I think his first crop is about 1/8 of the original image. His second crop is about 1/24 of the first crop. Therefore, cumulatively the last crop would be 1/192 of the original pic, that is about 0.5%.</p>

<p>I think the original 0.04% crop comment may be an unintentional mistake and Mauro did correct it to the 0.4% in a later post (page 2). I think Mauro's comment that comparing his 0.4% (or 0.5%) crop to a 2% crop from FF digital is similar to comparing a 35mm film scan to FF digital is reasonable because the crop difference is similar to the size difference between 67 and 35mm.</p>

<p>Mauro, thanks for the posting so those of us who still use film knows how much detail we can pull out of film. Currently I am still using 35mm a lot and sometimes I have problems with sharpness issue, this post shows me that the problem may likely lie within my overall workflow, instead of the film.</p>

Link to comment
Share on other sites

>> 377,300 / 3,795 = 99.4, or fractionally over 1%.

 

Given that the isolated eye crop that I posted on page 9 is about 1.1 percent of the entire frame, the eyes represent about

the same area of each frame. These examples represent a reasonable comparison of the two formats. Both show plenty

of detail. Use what you prefer and have no regrets.

Link to comment
Share on other sites

<blockquote>

<p>Use what you prefer and have no regrets.</p>

</blockquote>

<p>That about sums it up Dan. I won't argue that in terms of absolute detail, the 6x7 film will have much more real detail than the 5D2. That said, what I have found is that on normal large prints, 16x20, 24x30 for example....it doesn't make any real difference.</p>

Link to comment
Share on other sites

<p>Of course in the days of film most photos were taken with 35mm not MF, and today most photos are not taken with DSLRs but rather P&S digital or cell phone.</p>

<p>Having said that the photo does have a lot of detail in it, but for capturing memories is that level of detail needed? I mean to even start making good use of that level of detail you would need to make a print at least 20x30 inches and then view it from no more then 12 inches. I know a lot of people who really like photos of their kids but I don't know many that feel the need for photos that large that can be viewed that close.</p>

<p>I suspect that when future generation view those photos they are not going notice that great detail but rather the lack of color, but that is just my opinion because I like color.</p>

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now



×
×
  • Create New...