Using a DSLR for landscape photography

[robert goldstein]

Dave, what you say about scanning at 5400dpi makes sense, but the scans simply look better to me than those done at 4000dpi. Maybe others can explain it.

[godfrey]

Scanning at 5400ppi reduces grain aliasing (as I said up-thread) which helps the imaging

quality by improvements to the rendering of the grain.

 

Godfrey

[robert lee]

"... 5400dpi makes sense, but the scans simply look better to me than those done at 4000dpi."

 

Take the Velvia curve as a case study, where MTF response is down to 30% at 60 cycles/mm. You can look at this two ways (a glass half empty or half full thing.)

 

It's a bit nuanced. One can conclude that 135 format film sucks because a similar digital imager is still at 80% MTF in this region. Or, one can also correctly say that at least the film still _has_ a response: from here down to 0% MTF, the 6MP digital imager has no more pixels.

 

If the scene has high contrast, high detail features, 35mm film (especially negative film, and B&W like TMAX100 in particular) will still marginally outresolve 8-10MP DSLR. In which case, a higher resolution scan will dig out that detail.

 

Keep in mind what this means though in real life. A high contrast feature recorded at, say 5% MTF on film is barely discernable. It's also buried in tons of grain. Such might be interesting for forensics on surveillance images, but I'd hazard to say that it won't be of general utility.

[StuartMoxham]

Les I have looked at your scans many times now and have said often that while the film shows higher resolution than the 20D the lower contrast details are just not so visible in the film scans. Take the UC100 scans there are rivers that are almost missing the map grid lines are less visible. They are kind of lost in the grain. At the end of the day is does not matter because you would use not either format to reproduce a large map. Both formats would also be a poor choice if the intention is to make very large prints. Viewing 100% crops is like viewing a huge print. Now don't get me wrong I like film. Color Negative has great dynamic range. B&W films have there own look. If you make 8x12 to 10x15 inch prints you can get great results from both . Someone may prefer the smoother look of digital and others may prefer some film grain in a larger print. Just depends on who is looking and what the subject matter is.

[StuartMoxham]

Les I know your test is really about the resolution more than anything else. I also find the some of the pro digital claims a bit much. I have looked at the other scans too and of course RVP does a better job on the map than UC100. At the end of the day though people have to choose the right tool for the job and also what works for them. It is also important that they make the tests themselves using the materials and tools in the situation that they would normaly use them.

[dave_redmann]

Les and Robert Goldstein, I'll admit that my statement that scanning 35mm film at 5400 ppi is a waste of time is a bit of an over-simplification. To be more precise, I should have said that:

 

(1) There is no way you're going to resolve meaningful detail as fine as a 5400 ppi scanner might seem to give you. Just because a 5400 ppi scan of 35mm film has 39 MP in it does not mean you are getting anywhere near 39 MP worth of real detail, or anywhere near the detail that a 39 MP digital camera would give you. Fuji says Velvia 50 gives 50% MTF response at about 45 lp/mm and 30% response at about 60 lp/mm, which are about 2300 and 3000 ppi, respectively--and that's before you add in whatever falloff you have from the lens' MTF curve.

 

(2) You may well get marginal increases in resolution and/or sharpness by scanning at 5400 ppi, instead of lower resolutions. But I suspect that it has more to do with building in some extra performance in the scanner, in effect flatting its MTF curve where it counts, which for most cases is not much above 2500-3000 ppi. In other words, a 5400 ppi scanner may resolve 3000 ppi detail somewhat better than a 4000 ppi scanner does. But of course, the higher-res scan will certainly give you more grain.

 

(3) This is all in the context of color transparency film. Especially B&W is something of a different story--I'd be quite ready to accept that TMX can appreciably out-resolve Velvia, given the right subject, equipment, and technique. But of course, then you also get the issue of the silver grains scattering the scanner's light etc.

 

This would be an interesting experiment for making, say, 11x14 and 13x19 prints from 35mm transparency film: scan the film at 5400 ppi and reduce to 300 ppi to print, versus scan at 5400 ppi and scale down (bicubic) to 250, 200, and maybe even 150 ppi to print on a 300 ppi output device. I bet that if you use a printer run through a good RIP, you may well find that the scans you scaled down look better on the whole, and are very little if any less detailed, than the scans you kept at nearly full resolution. Grain should be reduced appreciably, but real detail should be reduced little, if any.

 

For the math challenged, you'd have a scan of about 5100x7650 pixels, which you'd scale down to:

3900x5850 pixels (300 ppi output from 4128 ppi effective scan),

3250x4875 pixels (250 ppi output from 3440 ppi effective scan),

2600x3900 pixels (200 ppi output from 2752 ppi effective scan), and

1950x2925 pixels (150 ppi output from 2064 ppi effective scan),

all before cropping, for the 13x19's;

 

and for the 11x14's:

3300x4950 pixels (300 ppi output from 3493 ppi effective scan),

2750x4125 pixels (250 ppi output from 2910 ppi effective scan),

2200x3300 pixels (200 ppi output from 2328 ppi effective scan), and

1650x2475 pixels (150 ppi output from 2064 ppi effective scan),

again, all before cropping.

 

Now output all of these files to a high-quality 300 ppi (or higher) printer with a good RIP.

 

I suspect that most of us would agree that the scaling the scan down to an effective scan resolution of around 2500-3000 ppi, if done properly (e.g., bicubic), involved little or no loss of visible detail in the print, and a visible improvement in grain / noise. This would be true even though the 13x19 would only have enough pixels to give you 200 ppi and the 11x14 would only have enough pixels to give you 250 ppi.

[robert lee]

"...marginalize the advantage of lossy desktop film scans of 35mm film and extol the advantages of digital..."

 

Les,

 

I don't see the contradiction between your film scans and the analysis of the published film specs. In fact, the scans illustrate the text quite well. Slow 35mm film does have more absolute resolution than an 8MP DSLR sensor, but it's not easy getting at that additional image information (and not without undertaking some fairly heroic signal processing efforts.)

 

Here's another way to think about this. The capture behavior of DSLR versus 135 film has its analogs in other arenas; for example, analog cell phones versus digital, FM versus satellite radio, etc. The digital implementation tends to be well behaved out to some hard bound, then falls off the cliff. The analog implementation degrades sooner, but remains usable longer.

[stan_belyaev]

Rob,

I don't see any reason to buy digital if you have a good film equipment. You will be sucked in an endless chase for a better resolution, software, computer. Use slides or negative film and enjoy.

I'm attaching a pic. of a tree. the original scan was about 20x30 inches at 300dpi (no interpolation). This is a tiny crop of the original 35mm image+low quality JPEG (due to 100kb limit)

Which digicam under 5000-8000USD can do that?

 

Les,

Thanks for your tests. Hopefully your site will be a revelation for some.<div></div>

[robert goldstein]

Stan, my feelings about Astia parallel yours, but I expect to see counterexamples from

digital camera fans.

 

It is hard for me to ignore the fact that a great many fine

photographers who excelled when using film now feel that they are producing even better

images with digital cameras. (I'm referring to 35mm film only. MF and LF are in a

different league altogether.) Can they all be deluded?

 

Rob

[stan_belyaev]

Rob,

I do have digital cameras (Contax N Digital and TVS Digital). Believe me Contax ND is a terrific camera if it is used properly. Disregard what people who have never seen or touched it say about this camera. Before them I had several other digicams. But every time I receive developed/printed negatives or slides from A&I (I use mailers) I stop thinking about digital. Well printed UC400 is beautiful. I experimented with local supermarkets and pharmacies and it didn't work. The pictures looked horrible. It appeares that there are only few places left which know how to process/print film correctly.

 

If you feel that your friends take better picture because of digital, then you should definitely try it. BUT you should buy the best camera and the best lenses (? FF Canon Mark...) (even if you have to borrow some money), otherwise you will regret (you will spend more money for upgrading you system in another year or two). I went through this.

If you r really serious about landscape photography then you may want to think about changing the format.....)

BR

Stan

[stan_belyaev]

Another example. This time it is Kodak UC400. Camera Conax TVSIII. Handheld.

Just a snapshot...

[stan_belyaev]

Another example. This time it is Kodak UC400. Camera Conax TVSIII. Handheld.

Just a snapshot... (JPEG qiality 50% due to size restrictions)

[lahuasteca]

Look at these links and make your evaluations. The D70s is with the 18-70 kit lens shot

at ISO 200, the F100 is with the 28-105 lens (we're pretty equal here) and using 400 UC.

The tripod is used with both. Notice particularly the leaves, water sworls, and wavelets in

the stream.

 

http://www.pbase.com/lahuasteca/image/69601322 (Bear Creek run with D70s)

http://www.pbase.com/lahuasteca/image/69601334 (Bear Creek run with F100)

http://www.pbase.com/lahuasteca/image/69601344 (Ohiopyle Falls with D70s)

http://www.pbase.com/lahuasteca/image/69601587 (Ohiopyle Falls with F100)

 

My own very unprofesdsional observation is that the F100/400UC resolved more minute

detail with the leaves and whitecaps in the water than did the D70s. I was suprised. I

used a Coolscan V at max. resolution, then rescaled for the web.

 

I still use the F100 for landscape where there is a lot of intricate vegatation; the D70s gets

utilized more for people and street photography.

[godfrey]

I hate "digital vs film" debates. As I said in my first post on this thread, <i>"Most of

today's DSLRs perform at least as well as 35mm film SLRs for landscape work, with

appropriate lenses and skills on the part of the photographer."</i> The equipment is

important, yes. The skill of the photographer in using it is at least as important.

<br><br>

Since the discussion has devolved to 'you can get more resolution from film than you can

from 6Mpixel digital' and several examples have been posted of film resolution, I thought I

would take the same image I presented there and extract a couple of detail clips at full

resolution from the original RAW file to give a practical example of what you can expect

from a 6Mpixel digital given sensible use of a tripod and a high quality lens. NO serious

landscape photographer would consider for a moment working hand held, in my opinion.

<br><br>

The scene in question was made on a dark, foggy, windy day so this does not represent

the ultimate resolution that this camera and lens can achieve. Rather it shows what a

typical capture consists of.. (My apologies if the composite image embedded here is a bit

large)

<br><br>

<center>

<img src="http://homepage.mac.com/godders/6Mpxl-DSLR-example.jpg"><br>

</center><br>

Note that white speck in the JPEG image that resolves well into a seagull flying over the

stormy ocean in the detail clip ... I hope the example proves useful for those unsure as to

whether a 10Mpixel camera can produce a

decent landscape image. 6Mpixel certainly does ok on detailing for 11x17 prints,

10Mpixel or more will give added advantages for cropping, detailing, and large print sizes.

<br><br>

Godfrey

[godfrey]

<center>

<a href="http://homepage.mac.com/ramarren/photo/PAW6/28.htm" target=new>

<img src="http://homepage.mac.com/ramarren/photo/PAW6/large/28.jpg" border=0>

</a><br>

Montpelier, Isle of Man 2006<br>

<i>©2006 by Godfrey DiGiorgi<br>

Pentax *ist DS + FA77/1.8 Limited<br>

ISO 200 @ f/13 @ 1/3 sec, Av (EV -0.7)</i><br>

</center><br>

<i>Les wrote:<br>

> regarding the "devolution of this discussion", it is usually<br>

> brought on by posting competent information on one side of it<br>

> then presenting completely unfounded opinions as fact on the<br>

> other. Take for example your first response above, you present an<br>

> image from your 10D (nice shot BTW) then state, "But much very<br>

> good work can be done with 35mm film, and better work can be done<br>

> with DSLRs in my opinion" so if your point is that a 6MP is<br>

> better then 35mm film, where is it? Does your statement mean that<br>

> this scene could not have been made with 35mm film?<br></i>

<br>

No, you're reading too much into it from your own biases. I meant exactly what I stated:

<i>"much very good work can be done with 35mm, and better work can be done with

DSLRs in my opinion"</i>.<br>

<br>

My opinions are based upon simple, clear, solid bases. I find it easier and more versatile to

work with digital capture than with film because:<br>

<br>

a) I can check exposure, framing, etc right on the spot without leaving the site. <br>

<br>

b) The image files are cleaner than 35mm scans and lend themselves to easier, better

quality image processing<br>

<br>

c) The dynamic range of digital capture with RAW format surpasses the majority of films

available in my experience, making it possible to capture scenes that are vexing to acquire

with film. <br>

<br>

d) Working with digital capture is less time consuming and requires fewer steps in

process, while producing comparable results, compared to working with 35mm film and

producing prints, regardless of which specific capture to print process you desire to

pursue. <br>

<br>

Note that my opinion is based not on resolution requirements, which are only significant

with respect to how big a print you want to produce, and are all about getting the job

done in a practical sense. <br>

<br>

<i>

> Look, resolution is by far the easiest characteristic to prove -<br>

> just conduct the test as I outlined in my Digital and Film<br>

> Relative Resolution Album as it is independently verifiable and<br>

> you'll see where your setup compares to the various desktop scans<br>

> of 35mm films that I have already posted. Since I already posted<br>

> my 8MP Canon 20D in it, I can unequivocally state how much detail<br>

> it can resolve relative to the others.<br></i>

<br>

I've been to your pages many time and find the information there both difficult to

understand as to your testing procedure and what the results show, as well as irrelevant to

my work. I don't disparage your efforts, I just can't make out how it is relevant to me as I

find sufficient detail resolution in practical use with my DSLR to make the prints I want at

the quality I'm after. (And yes, I have 6, 8 and 10 Mpixel digital cameras to work with; I'm

not stuck on some absurd notion that a 6Mpixel camera is the be-all and end-all...)<br>

<br>

I find that 6Mpixel digital capture provides sufficient resolution for A3 and sometimes A3

Super sized prints. 10Mpixel capture will allow an increase in the size output by up to

25-30% linear magnification. This is sufficient for the vast majority of landscape work that

I'm interested in, or see presented in gallery exhibitions. <br>

<br>

I do still enjoy photography captured with film media. I feel that it has a different feel in

many cases but, moreover, it invites a very different psychological approach on the part of

the photographer. As a matter of fact, I've recently invested into some medium format film

equipment again because I want to explore that psychological difference in the experience

of creating the photographs ... although, for all practical purposes, I feel the use of a

Canon 1Ds II or larger format digital cameras obviate the need for medium format on the

basis of resolution and quality, and certainly with respect to workflow and productivity.

<br>

<br>

Godfrey<br>

[rob_landry]

I own a Minolta 5400 scanner along with a 6MP DSLR and I find that my film scans yield more pleasing images than the same images shot with my DSLR. Granted, the differences are not massive, but still noticeable. In addition to somewhat higher resolving power, I find that my film scans have far fewer artifacts and better colors. As far as noise goes, the film scans have lower shadow noise (a benefit of the scanner's multiscanning), and look as clean in other areas as the digital shots when printed. I run all of my film scans through Noise Ninja with excellent results. The massive amount of pixels allows me to use noise reduction without destroying detail and since I have to downsample to print, any remaining noise is averaged out. The digital shots on the other hand have to be upsampled to print big, so any artifacts that are present get magnified.

 

I also find that some of my DSLR shots suffer from purple fringing along high-contrast areas more than I would like (never got this with film), and with very fine detail, you sometimes will get strange patterns or artifacts where the camera was unable to resolve detail and made a "best guess". Film, faced with the same detail, will resolve, albeit with lower contrast but this appears much more natural. Film also handles highlights (reflections on water for example) with much more grace than my DSLR.

 

With all of that said, I can't say that I enjoy either format for printing landscape images too much beyond 8x12. If you enjoy landscape photography, you may want to consider at least 6x7 MF or higher. Personally, I've been shooting LF for quite a few years now and wouldn't consider using anything but for landscape work. The results are spectacular and using a LF camera is a very enjoyable and rewarding experience.

[stan_belyaev]

I read an interview with the Fuji Reala's developers on russian fujifilm website. It was published in 2003 (russian only)

Those guys mentioned that a regular color film (35mm) has about 20mil. picture elements, Reala has 50 millions.

I don't know whether an information could be extracted from each of the elements, but the numbers sound very impressive.

[rob_landry]

Les,

 

I appreciate your effort to provide a comparison, however, it would be nice if you were to include the details of each shot and scan settings. Also, in reviewing your film scans, they do not appear to have much, if any sharpening applied. This of course is not realistic as all scans are meant to be sharpened. DSLR files undergo significant processing before they are output as a file. Also, with regards to the DSLR shots, are you remembering to alter the focal length or shooting distance to allow for the FOV difference?

 

Again, thanks for undertaking the task.

[godfrey]

Les,

 

Bias used in the context of what I've said means that you have stated your opinions and

presented your take on supporting data. It says nothing about demeaning the opposite

opinion.

 

I have not said one way or another that film is or isn't higher resolution. What I've said is

that in my opinion the DSLR produces adequate quality to match and eclipse the use of

35mm film for landscape photography. I've stated this based both on the several pictures

I've shown here (and the many others available for viewing on my website), with the 20+

years that I've spent doing digital

imaging, and the 40+ years that I've been doing photography of all kinds as background.

 

I do not choose to engage you in a semantic debate that has no purpose. What I said is

objective, neutral, and represents the information at my disposal coupled with the

experience I have

of doing this work.

 

Godfrey

[godfrey]

Les:<br>

<i>... "it's simple - bias is you only presented favorable things about one side while

berating the other without so much as a reference or proof of what you believe to be true."

...</i>

<br><br>

A: You're wrong, and what you have stated is untrue. <br>

B: I don't need to provide a "reference of proof" to state my opinions. <br>

<br>

Good night, and good luck.

<br><br>

Godfrey

[roger_smith4]

I read this whole miserable thread. Les, you're being unreasonable here. Godfrey is saying that DSLRs are technically good enough and workflow benefits are significant enough to make DSLRs preferable for landscape use. That's his opinion, and not something provable or disprovable. You can certainly disagree on the basis of resolution or what have you. I think you're probably both right as you're talking about different things.

[godfrey]

Robert Goldstein should be the one to judge whether or not his original question has been

appropriately addressed and by whom, Les. I've offered my information and opinions to

address his question, offered examples of work in way of addressing that question.

 

So far, all you have posted on this is your opinion based on a set of amateur resolution

tests. Do you have any landscape photography you would like to show which demonstrates

your thesis?

 

BTW: If you continue to pursue a stupid flame war, I will not participate in it.

 

Godfrey

[dave_redmann]

<I>I find that my film scans have far fewer artifacts and better colors.</I><P>

 

What are better colors? That's a totally subjective standard. Good DSLR's have much more <I>accurate</I> colors than any film. Now if your personal preference is for a film's colors, that's fine, but it means you prefer less accurate colors (which may indeed be preferable, in some cases). As to fewer artifacts, do you shoot JPEG's, or convert from raw to TIFF?<P>

 

<I>Those guys mentioned that a regular color film (35mm) has about 20mil. picture elements, Reala has 50 millions.</I><P>

 

Fuji's data sheet for Superia Reala 100 shows the MTF curve falling below 50% response at about 60 lp/mm, which would correspond to a resolution of about 12 MP, assuming the lens was perfect out to that resolution. Even using the much more liberal standard of 30% response, you still only get about 75 lp/mm / 19 MP. The curve does show <I>some </I> response out to about 110 lp/mm, but at that point you have almost no contrast--this is not really usable resolution, and film grain will mostly mask it anyway. Just FYI, remember when viewing these curves that the scale is logarithmic.

[robert goldstein]

<Robert Goldstein should be the one to judge whether or not his original question has been appropriately addressed and by whom>

 

Right you are, Godfrey, and here is what I have learned from this discussion. There seems to be considerable disagreement regarding relatively minor differences between the two modes. I take this to mean that there is no clear cut winner and that I will have to find out for myself if digital capture produces pleasing results in my hands. Much of the discussion has revolved around 6MP cameras, and I feel confident that the latest generation of 10MP cameras are capable of rendering significantly more detail. If they can equal or surpass high resolution scans of Astia, I will be satisfied. Now, whether I will enjoy lugging around a large SLR kit is another question altogether.

[godfrey]

<center>

<a href="http://homepage.mac.com/ramarren/photo/PAW5/large/42O2-half.jpg"

target=new>

<img src="http://homepage.mac.com/ramarren/photo/PAW5/large/42O2.jpg">

</a><br>

In Lighthouse Park - Santa Cruz, California<br>

<i>©2005 by Godfrey DiGiorgi<br>

Konica Minolta A2<br>

ISO 100 @ f/3.2 @ 1/60 sec<br>

Click on image above for larger rendering. </i><br>

</center><br>

Very sensible, Robert. You will find out best what works for your photography that way. I

hope my comments have been helpful to you.

<br><br>

You mention wondering whether you're going to enjoy carrying around a "large SLR kit".

I'm curious as to what kind of film camera you're carrying now as a basis of comparison. A

Pentax K10D 10Mpixel body with two or three lenses is comparable to my Nikon FM/FE2

bodies with the same lens kit for FoV in overall size and weight.

<br><br>

Godfrey