large format versus 10MP dslr at web sizes

<p>I was having a discussion with someone who shoots with a 10MP dslr as well as a large format camera. He is enjoying the huge resolution increase and overall look of film, and began talking about the need to upgrade his dslr to a newer model. One reason he gave was that in viewing low res shots on the web (like flickr) he could see big improvements in resolution from other cameras over his 40d. He also said he noticed that there was more detail in shots at low res, 500-700 pixels, from his large format than his current camera. I believe at small sizes, 700 pixels or so, sharpening methods are much more important for the appearance of detail than the resolution of the cameras one uses. Ergo, at that size, you won't be able to tell a significant difference in resolution at that size between say a 10 MP camera and a 36 MP camera, or even large format camera. And any appearance of enhanced detail between film and digital at such low res probably boils down to the way film handles contrast, etc and not due to increased resolution. Therefore, you need to look at larger file sizes or close-up crops making a valid visual comparison.<br>

Do you concur? Why or why not? Thanks in advance. </p>

<p> </p>

<p>Not sure if I'm addressing your concern...and if not, try to ask in a different fashion.</p>

<p>There was a test done last Summer (all things equal...supposedly) between D800 and IQ180...yes, $43,000 sensor. After the tests 40x60 prints were made and they were nearly impossible to determine which print was better.....from same viewing distance. They had certain nuances, but nothing drastic. What I'm saying is that this was closer "apples to apples" comparison. Till one prints the shots to certain size, they all will usually look superb (my 12MP rig included) on the screen. By the way I have the pics (comparisons), but for legal reasons I prefer not to republish them on the net. </p>

<p>Film has it's own qualities, but unless those qualities are translated properly in a scan, too often lots of info is lost. </p>

<p>JPEGs tend to degrade the image at some point. IMO, if I keep it above say 600, it would be difficult to differentiate from the original RAW file. I'm talking average. LF has more resolution and most 'togs are aware of this...and that also reflects in the lower resolutions. Both, the digital and the film require certain pp approaches.</p>

<p>Les</p>

<p> </p>

<p>10 MP is more than enough for 600-800 PX "web" resolution. There may be other things your friend notices with other pictures, but native resolution in the sensor is not one of them.<br>

1. It simply could be the images are "cleaner" due to the improvement in how the newer sensor records light, no that it has more MP's. It is nearly impossible to tell the difference between a Nikon D3s 12 MP and the DXs 24 MP. To say you could see a difference at web resolution is nonsensical.<br>

2. I shoot with a Canon 30D. Prints that have been made with Zeiss glass look much better due to the better resolution the lens place's on the sensor. Like film, the sensor is only as good as the light that is placed on it.<br>

3. His particular 40D may have an issue as well. <br>

More than likely, the IQ improves due to the increased quality of the pixel in the newer sensors, not the overall <em>increase</em> in pixel count. The surface area to collect light remains the same. The increase in pixel count is more a factor of <em>marketing necessity</em> than of image quality.</p>

<p>His camera is three-generations old. Differences in pixel density are unlikely to show on Flickr unless he allows viewers to view his "Original" size files (an option on Flickr). The main gain will be in in-camera processing and sensor technology, giving better DR and high-ISO performance, which can be apparent in sizes as small as 600 pixels. For many subjects, these improvements will not be apparent in internet sized images, but in a few, they'll be apparent.</p>

<p>I regularly view my Flickr Contact's images at 1024x683 pixel sizes. I also view in the Slideshow mode that fills my 24" monitor. At these sizes, particularly with cropped images, then the pixel density and sensor performance becomes noticeable. Many Flickr users do take advantage of these viewing alternatives. My Previews are 400 pixels.</p>

<p>A friend of mine does comercial and portrate work with a Nikon D800. He shoots jpegs at whatever pixel count that camera gets, pretty big I understand. The advantage to him in his work, is that he can view each member of a group shot as an indevidual headand shoulders portrate to check for closed eyes and so on. I have seen him swaping eyeballs to get an overall great group image (no, its not ART).</p>

<p>Don't know if I could scan a 4x5 transparency to that resolution and manipulate it like that, but then I don't want to and don't do that kind of work.</p>

<p>Yes you could and no, I wouldn't want to either.</p>

<p>if a high resolution digital is 10 or even 20mp<br>

and a 35mm full frame shot on c-41 or other film is 20-22 mp eqyuvalent<br>

what is the mp equivalent of a 4 x 5 or 8 x 10<br>

negative or transparency ?</p>

<p>I'm not going to do the math, but you're talking about a multi-image stitched digital image to equal an 8 x 10. People are actually doing that and ending up with multi GB files. I've done some digital panoramas and ended up with compressed jpeg over 100MB and the TIFFs were approaching a GB.</p>

<p>Of course, all this discussion ignores the inherent differences of view cameras vs. box cameras. i.e. perspective control etc.</p>

<p ><a name="00bLOz"></a><a href="/photodb/user?user_id=2074460">Mike Bischof</a>, said:</p>

 

<blockquote>

<p>Of course, all this discussion ignores the inherent differences of view cameras vs. box cameras. i.e. perspective control etc.</p>

</blockquote>

<p>Well, of course, since we're talking about film vs. digital, then it should be acknowledged that software can control perspective, etc.</p>

 

Software can't "control perspective, etc.". It can distort an image to make it look as if perspective has been controlled, but not quite.<br>It's convenient, compared to using movements of a LF camera, yes. But not the same, not as good.

<p>I have (and regularly use) a 4x5, a 6x6, a 24mp DSLR, and occasionally a 16mp DSLR or 35mm film.</p>

<p>I can honestly say that downsized to web resolution, the ONLY advantage of film, even 4x5 film, is that depending on the image, it may have a wider tonal range than a DSLR. Some new full-frame DSLRs rival or beat some medium and large format film stocks, but absolutely NONE of the older models do.</p>

<p>Now web sharpening is a wholly different beast. Because of the way downsizing and scanning vs. digital capture works, some formats just plain look better on the web, for no other reason than pure coincidence of design. I would recommend trying and/or buying a copy of the PhotoKitSharpener plugin for Photoshop. It has output sharpeners for print or web/mobile use, and operation is basically automatic. I've found that every single time I've used it, my photos look better on the web and mobile devices, and <em>most</em> times my prints look better too. It's <em>never</em> made my prints look worse.</p>

<p>It's a hundred bucks, but it is money VERY well spent. I can honestly say that it made a definite and perceptible, if small, improvement on almost every photo I've used it with.</p>

<p ><a name="00bLVY"></a><a href="/photodb/user?user_id=282122">Q.G. de Bakker</a> <a href="/member-status-icons"><img title="Hero" src="/v3graphics/member-status-icons/hero.gif" alt="" /><img title="Frequent poster" src="/v3graphics/member-status-icons/3rolls.gif" alt="" /></a>, said:</p>

 

<blockquote>

<p>Software can't "control perspective, etc.". It can distort an image to make it look as if perspective has been controlled, but not quite. It's convenient, compared to using movements of a LF camera, yes. But not the same, not as good.</p>

</blockquote>

<p>If you're not happy with your software, then buy a TS lens. You may not be able to make as extreme adjustments as with a LF camera, but within reason, you will gain tons of control.</p>

 

You forget what forum we are in, David? I doubt anyone here needs a TS lens when they are not happy with their software.<br>There are, of course, more advantages to LF then perspective control. But i can't see any as long as we are talking about images for the web.

<p>I appreciate the discussion and the input this has generated (even though it did wander off topic in a few posts), and thanks to those who addressed my question. Zack, it was interesting that you say some newer full format dslr's can beat the film stocks. Which films are you referring to and which parameters are you thinking of?<br>

Leszek, I'm interested in the test you mention, but based on your response, it sounds like the digital files from which the prints were produced are not available online or for public view. Is that right?<br>

David, do you have any example shots you can point to which illustrate the differences? One hindrance to making conclusions with flickr views between users is the lack of post-processing data...however, I think it may be possible to make some hypotheses based on patterns over a wide range of pictures. </p>

<p>Jay, you can use <a href="http://www.pixel-peeper.com">www.pixel-peeper.com</a> to find some images taken with a 40D and others taken with, for instance, a D800 or a 1DX or a 5D MkIII. I'd suggest searching for images at ISO 800 and then looking at them in the 2048 size, which will fill most monitors. They were likely taken by different people, at different places and processed differently, as you say, but if you look at a handful of each, you'll see that the 40D, in general, is not up to the latest generation.</p>

<p>Regulars on Flickr often do look at each others' pix full-screen. The definition of "internet sizes" is changing. For use as a non-featured background image on a web site, yes, you can't tell the difference; however, on photo sharing sites, you can see differences. As monitors improve, with pixel counts approaching 5,000, it'll be even more noticeable. Those monitors are available now, but outrageously expensive, so it may be two or three years before you and I consider one. Still, it is coming.</p>

<p>BTW, if EXIF is shared by the account holder, you can see what software was used, if any. In-camera JPGs will not show a software. Not everyone shares EXIF data.</p>

<p>Q.G. we're in the LF, Digital Forum and the OP wanted to compare LF to a 10MP dslr digital image for internet usage. I pointed out that internet usage is changing, with some sites allowing larger and larger files, but that small files wouldn't show a difference.</p>

<p>Since dslr was introduced by the OP, then a TS lens is appropriate for the discussion, if someone is concerned about perspective and doesn't think that their software is good enough.</p>

 

It's never been the case that my MF or LF scans look bad when down-rezzed for web use, detail is always stunning if it's there in the original file. It's frequently a minor challenge to get a small downsampled 10 or 12 MP DSLR file sharpened at 600x900 without blowing out contrast or subtleties. (But it really doesn't take more than moments to go into Photoshop and manually work with the USM settings for the DSLR images... nothing on the order of working with LF film and scanning it in the first place.)

<p>I use an ancient 10meg Sony A100. I use it for researching subjects for MF. Look at this thread in Classic Manual Cameras: "50 years of Linhof". 10meg Jpeg straight out of camera is more than adequate for my "web" needs. They even look good on a 32 inch LG 1080 TV.</p>

<p>However, I must give up on equipment "creep" and NOT trade the Sony A100 for an A850 or heaven forbid an A99. But I say to myself that the 100mm f/2.8 macro is going to waste! And I don't use the Maxxum 9 anymore. Sigh!</p>

<p>A bit earlier in this thread there was the view that perspective correction in Photoshop was not "the real thing", that is, the result would differ in shape from taking a picture with a view camera, with the lens shifted in parallel vis-à-vis the film holder. </p>

<p>I tend to disagree with this view. If done in the right way, the result is exactly the same (you should start with 16 bit precision though, since parallel correction in Photoshop destroys some pixels).</p>

<p>What is the right way of doing perspective correction in Photoshop? Just drawing the upper corners of the picture outward to correct for "falling" houses does not do the trick. You have to divide up the movement into two. Half of the correction should be done by drawing the upper corners outward and the other half by PUSHING THE LOWER CORNERS INWARD. I challenge you to see the difference between a picture thus corrected in Photoshop and a picture taken with a view camera lens with parallel shift.</p>

 

Emil,<br>And why would you 'push in' the bottom halve, instead of drawing out the upper all the way?<br>But however that maybe: what either does is distort the image in one direction (sideways), not in the other (hence "distort"). And there you have a first difference between using movements and using post-processing software. ;-)

<p>Dear Q.G.,<br>

It appears to me that you have not yet tried to digitally correct the "falling back" effect of pointing the camera upwards. If you just draw the two upper corners apart you will distort the picture, with the house in question becoming unnaturally broad. By using the method I outlined earlier, the apparition of the house will be preserved and you will get an outcome identical to that obtained by shifting the lens upwards vis-à-vis the film holder, except a small difference arising from the fact that the view camera lens has a slightly elevated viewpoint. The latter effect would be countered by raising yourself on your toes when taking the picture with your digital camera ;-). </p>

Emil,<br>If it would appear so, it would be deceiving. I hav been doing it for longer than i care to remember.<br>If you draw the top apart a bit, the bottom in a bit, you distort the building the same as when you just draw out the top (or the bottom in). The shape the image ends up in is, has to be, the same if the builing in it is supposed to look the same. The only difference is that when drawing out the top only, you're not throwing away image data in the bottom half of the picture.<br><br>The difference with using movements is - as said before - that doing this in PS, you are distorting the image in one direction only: sideways. You are doing such in a differing degree from the bottom (not at all when you don't, unnecessarily, pinch the bottom), to the top (most). That is not what happens when you use movements. The outcome is not identical.<br>(Mind you: i'm not saying that using movements leaves the image undistorted. Any view of a building that has converging verticals 'corrected' is a distortion.)<br><br>Have you used a camera with movements, or a shift lens? Maybe the latter. I ask, because you assume that the lens <i>"has a slightly elevated viewpoint"</i>. Something users of view cameras know not to be so, because they do not change a selected viewpoint just because they want to correct converging verticals. The back drops... ;-)

<p>The biggest obersvable difference in resolution between your 10Mp DSLR and LF will be down to lens rendering, especially with wide angle lenses. Quite a few lenses smear two or even three pixels which can reduce the perceived resolution in corners by 75% or more. i.e. it's visible in a 800px by 600px thumbnail. LF is pixel perfect in the corners at 100% for most lenses and at 10% you can't differentiate. <br>

However, I think a lot of the difference may be down to sharpening differences (resizing algorithm changes). <br>

p.s. there is no way a D800 at 60x40 would look indistinguishable from an IQ180 file. Here's a picture of a target taken by both at the same distance - the shots were taken on different days so the hasselblad camera isn't at the same angle. <br>

http://static.timparkin.co.uk/static/tmp/iq180-d800.jpg<br>

This image has been scaled to 300dpi at 60"x40"</p><div></div>