The Online Photographer

Check out our new site at www.theonlinephotographer.com!

Friday, July 28, 2006

'Meeting' and 'Beating': Relative Terms

I'd think I'd probably better clarify this perhaps overly dramatic statement I made in the previous posting:

"In B&W mode, it's the first digital compact that meets and beats [35mm] Tri-X."

Well, maybe. I shoot 35mm Tri-X using in-camera metering with the exposure index set according to the light: E.I. 200 for most subjects and general shooting, E.I. 100 for contrasty (usually bright) light, and E.I. 400 for flat (usually low) lighting conditions. I develop it all the same, 8.5 minutes in D-76 1+1, and control print contrast (naturally) with contrast filters and VC paper. (I detailed this further in one of my columns a few years ago in Black & White Photography magazine, and if I could find the issue I'd tell you which one it is. Have I ever mentioned that I'm not the most organized person in the world?) It's a practical, utilitarian system that yields high-quality negatives in most lighting, while not requiring the hassle and labor of developing individual filmstrips for specific shooting conditions, which is sort of practical with 12-exp. 120 film but almost always a great annoyance with 35mm unless you're working with three camera bodies (this puts me in mind of Dan Weiner, who did exactly that...).

Enlarged using a high-quality enlarging lens and a diffuse light head enlarger I'm almost always happy with enlargements of 7 diameters (7x10.5") and sometimes 8. I would say I get up to 8 stops of subject luminance reliably covered in the prints. I could probably claim 9 and get away with it.

(Damn, as I write this I'm already seeing that a rigorous comparison between 35mm Tri-X and the F30 would be educational for me and informative for others. Another set of trials I really don't have the time to carry out.)

Previously, the best small-sensor digicam I owned was the Sony F-707, a really nifty camera that yielded barely adequate results in color and was essentially useless for B&W except under controlled or ideal conditions.


Here's one of the best low-light pictures I made with that camera. It was shot at ƒ/2 and 1/30th at E.I. 320. It has worse "grain" than Kodak P3200 shot at 1000 and probably no more than—I dunno, what would you guess? Three stops of range, maybe four, about like slide film. Way too little for the picture to work in B&W, at any rate.

So I've been using the little F30 in B&W with compensation set on –2/3 or –1 to help hold highlights. If you go back to my F30 shots posted the other day, you can see that the "highlight" of the sunlit macadam in the background of the first shot is totally blown. The second shot, below, makes DR easier to guess—it was a hot, hazy/sunny but bright day. Again, I didn't meter anything, so I'd have to guess based on experience, but what would you say that's giving me? Five stops with tone and detail? Maybe as much as seven. (I'd be interested to hear Carl's take on this, as he's particularly good with this sort of thing.)


So anyway. The F30 beats Tri-X for speed, which is a first in a digicam for me. It approaches it in DR, although that's about all. Tri-X has latitude for additional underdevelopment, but digital files have the nuclear option of the Shadow/Highlight control in Photoshop, especially in B&W where chroma noise isn't so much of a limitation in the shadows as it is in color. Finally, 6 megpixels prints pretty well at 7x10.5".

Posted by: MIKE JOHNSTON

16 Comments:

Blogger Luis said...

Every picture lives in its own world, but usually shadow/highlight imho is a quick passport to achieve obviously crafted prints.

2:13 AM  
Blogger sergio said...

Why would you want to shoot at different ISOs if you develop everything with the same conditions? It is exactly the same as giving more or less exposure.
What's the point in doing that? Am I missing something?

7:40 AM  
Blogger Photo-essayist said...

Mike, I know you have a great deal of experience printing B&W, but aren't you being a bit conservative printing Tri-X only to 7 to 8 diameters? I love the quality of Tri-X grain and usually want to see the print pushed out a a little further where one can taste the grain. With a glass carrier and an excellent enlarging lens, sharply resolved grain gives the print overall a convincing shaprness. Of course, this is old hat to you. For my money, I'd want to compare the output of the F30 to a larger print from Tri-X, perhaps printed to an 11x17 sheet. Just one guy's opinion.

7:40 AM  
Blogger Mike Johnston said...

Photo-essayist,
As you know, the issues of enlargement are going to be different. Bigger prints from digital files simply look different than bigger prints from film. Comparisons aren't going to tell you a whle lot except about the limits of your taste for each technique.

--Mike

7:49 AM  
Blogger Mike Johnston said...

Sergio
Yes, it's giving more or less exposure based on the lighting conditions. As I said, I use the camera's autoexposure metering, so I'm not setting the exposure myself except by adjusting the ISO.

--Mike

7:51 AM  
Blogger Chris said...

Just for fun, I decided to test process the color image you mentioned as being unsuited for b&w, and I got a reasonable result from your low-res source (that wasn't the original source file dimensions, right? If it was, then, yeah, you're right, it's not at a quality for larger printing). I'd be curious to see how the original source faired under the same prep.

I converted to 16bit, and used three manipulations: LCE (using unsharp mask), a b&w conversion using a color filtering process (two h/s adj. layers, the lower one set to "color" mode, the upper layer set to normal, but saturation at -100%. Adjust the lower h/s layer to filter the image), and a curve at the end.

(This was prepped on my laptop screen, so all appropriate caveats apply.)

Here's the image.

8:58 AM  
Blogger Just Plain Hugh said...

E.I. 200 for most subjects and general shooting, E.I. 100 for contrasty (usually bright) light, and E.I. 400 for flat (usually low) lighting conditions. I develop it all the same, 8.5 minutes in D-76 1+1, and control print contrast (naturally) with contrast filters and VC paper.

Sounds to me like Expose for the shadows , print for the highlights

I used to do the same thing, shooting Pan-X 120 at between 20 and 100 ASA , and developing in D-19 1+1 to get negatives that had an incredibly long tonal range . I used a diffusion enlarger , Portriga Rapid, and LPD for printing.

The adjacency effect of the diluted D19 was the key to the whole thing.

Photoshop's shadow/highlight and unsharp mask are sort of analogous to the adjacency effect, but the dynamic range of digital camera sensors and their linearity compared to the H=D curve of film leave a lot to be desired

9:06 AM  
Blogger Mike Johnston said...

Chris,
I thought someone might try that. (s)

That's a good conversion, considering, but tonally it looks about like a halftone on newsprint to me. The problem is that the original file recorded maybe 4 stops of DR. An ordinary B&W print might have 7 stops, and a contrasty situation like that one would look best with 9 or 10 stops covered--a fairly severe "pull" in film terms.

--Mike

9:22 AM  
Blogger Dwight Jones said...

A number of people have commented that the Ricoh GR looks like Tri-X when used at ISO 1600. Some of the same people gripe about the lack of a fast lens. It seems to me that we're already starting out with a two stop advantage, so the f/2.8 lens should be just fine unless we're going for selective focus. Similar arguments could be made for the F30 or any other digicam. The piont is that no digicam is going to be exactly like film. That's OK if we enjoy the strengths of each system.

9:24 AM  
Blogger MarkT said...

I tested the Ricoh's grain characteristics in response to the earlier thread on photo.net mentioned by MJ (last week?). A lot of people say it looks like Tri-X. I posted some results here if you are interested. The other interesting point was raised that you can get markedly different looks depending on whether you shoot RAW or use the in-camera Jpeg engine.

9:43 AM  
Blogger Chris said...

Mike,

If possible, could you translate your comments regarding DR and such into digital-related terms? Things like "stops of coverage" and "severe pull" have no meaning to me, since I never worked with film or film printing, so I'm sitting here a bit confused. (Or, if you know of an online resource that is useful for this, it would be greatly appreciated.)

For example, 4 stops covered, 7 stops covered -- does that have a translation when looking at a histogram of an image? I'm looking at the histogram of the larger version of that color image, and it seems to cover the full tonal range, from shadow to highlight. This is, I assume, due to significant post-processing on your part?

Is it the grain of the image that gives you pause? Or the contrast of the grains within certain spaces? Is it the introduction of banding due to large tonal adjustments?

If you could go into more detail, I'd appreciate it. I find that b&w digital is very flexible, and large adjustments of tonal range can still lead to great results, if done right, so I'm curious to know what features you are specifically looking for.

Thanks!

10:04 AM  
Blogger Mike Johnston said...

Chris,
A "stop" is just a doubling or halving of the light, 2X or 1/2X. In photographic negative terms (at least using Phil Davis's BTZS terminology), there's "DR" meaning density range on the negative, and "SBR" meaning subject brightness range. The idea of exposure and contrast control is to fit the one to the other optimally.

You need to separate, in your thinking, the idea of the tonal range of the image from the tonal range of the subject. The problem is not getting a range of tones from black to white in the image--that's easily done with inadequate dynamic range--the problem is converting the subject's luminances to image luminances such that there's tonal separation (texture and detail) where you were able to see it in the scene, or where your eye expects to see it in the image.

So, for instance, if you look at your B&W conversion (I'm assuming you're the same "chris" who wrote the earlier comment), notice where you see nothing but dead black in the image: the bottom of the salad bowl, in parts of the towl, the shaded area of the basket, the area behind the door, etc. Then look at the dead white highlights, for instance around the area of the towel. The sensor has just not been able to capture all the luminances present in the original scene.

A histogram of a scene entirely within the DR of the sensor would be one that appears to be a "mountain," i.e., that starts and ends within the range of the graph. You can then match up the scene DR and the image DR perfectly by setting your black and white sliders to either side of the "mountain." (I don't really know how to describe this in words.) This one would be the opposite, where both sides of the histogram butt up against the edge of the graphic, showing that you're losing information at both ends.

Is this clear at all, or just more confusing?

--Mike

1:37 PM  
Blogger sergio said...

Hi Mike, what I meant with my post is that contrast is controlled by varying development times rather than adjusting exposure.

2:26 PM  
Blogger Chris said...

Mike,

Yeah, I understand what you're getting at. Thanks for the clarification!

But your latest comment also points out the difficulty in having a conversation about any arbitrary image via the Internet: your impressions of the b&w image didn't jibe at all with my recollection of it, and they sent me back to the image to take another look. The shadow and highlight detail not seen in your particular configuration is visible on my laptop (and if I viewed it on the CRT of my desktop, it would surely be different there, as well). The salad bowl, for example shows a smooth gradation of tone (well, smooth enough) from top to bottom, and only a little clipping under what I assume are the salad tongs.

An eval in photoshop shows very limited clipping at either end, mainly in the specular highlights and lamps, and in the core of certain shadows, and referring back to my photoshop document, the clipping seems due more to my final adjustment curve than to a problem with the original image. Although the original color image shows significant highlight clipping of the red channel, and almost complete shadow clipping in the blue channel, once the b&w conversion is applied that clipping disappears (mainly because those channels get dialed back significantly in the "color filter" conversion).

I think, when it comes down to it, we would have to be side-by-side, looking at a print of the image, or a trusted monitor (as far as we can "trust" a monitor) to see how each other assesses the same image.

And although monitor calibration can go a long way toward ameliorating these variations, unless we're using the same calibrator, and our monitors have similar dynamic range capabilities, it still only be a half-way solution.

Thanks again for the clarification, and thank you for the great commentary (and your active participation in your comment section).

2:32 PM  
Blogger Mike Johnston said...

Sergio,
Not really. Or, rather, that's one way, if you use sheet film. But actually the system developed by C.E.K. Mees and Loyd Jones in the 1930s was to develop the film to an average C.I. (contrast index) and then control contrast with paper grades, and that's pretty much what small format photographers have been doing ever since.

--Mike

2:49 PM  
Blogger Mike Johnston said...

Chris,
Some valid comments, but again, you're looking at the image, not the scene. If you could measure the scene, you'd find a lot more range than exists in the image.

--Mike

2:57 PM  

Post a Comment

<< Home