Velvia 50 vs.

When I was 10, I very excitedly read a book on 35mm photography from cover-to-cover. It didn't talk about slide vs. print film except to say that they're developed using different processes, and for a long time I thought slide film was a novelty for people who wanted to project vacation slides.

It was only a few months ago, when I first read Ken Rockwell's glowing recommendation of Velvia, that I began to change my impression. Over Thanksgiving, I flipped through the photography book again and noticed that all of its photos were taken on slide film! So I had to try it to see for myself. Here's what I found:

Velvia vs. itself

One of my biggest worries about slide film was its exposure latitude. I'd heard that you have to be dead on with your exposure and bracket any shots you really cared about.

After shooting a roll, I think that's an exaggeration. I didn't want to expend film on a real bracketing test, but here's one that happened accidentally. Even with automatic matrix metering, pointing the camera at different locations produced different exposure settings, so I took two shots. There's not much dynamic range to begin with (overcast day), but there's a 1 stop exposure difference between these two shots, and there's not a huge difference.

Velvia

Velvia, +1 stop

In this case, the folks who scanned it did some exposure tweaking: the difference is slightly more apparent on the original slides. My exposure log might also be off.

When faced with a scene with a larger dynamic range, you do have to choose what you want well-exposed, and the auto matrix meter might not do what you want. You could use spot metering; my technique is to take a picture on a digital camera (I used my iPhone) and use that to decide how many stops to adust the compensation on the auto-exposure. It also doubles as a date/time/GPS-tagged index print!

iPhone metering shot

In this case, I set the 35mm camera's exposure compensation to -1 stop. But the longer lens cropped out the darker areas, so this shot ended up -2.8 stops relative to the iPhone shot.

Velvia, -2.8 stops from iPhone

Almost 3 stops, and it looks fine. So it seems there's still a bit of tolerance with exposure settings. Not so scary.

Things are going to be contrast-y in direct sunlight, but I expect that. In order to fit what you can see (human eye: huge dynamic range) onto what you can display (prints or monitors: small dynamic range), there must either be clipping (slide film) or dynamic range compression (print film, or HDR in a more extreme case). Compression preserves more detail and may be more appropriate if we develop high dynamic range displays, but currently I think clipping is the more naturalistic option.

Of course, this begs the question: can post-processing simulate the effects of slide film? I think the answer is yes, but only if you're careful (more careful than I care to be).

Velvia vs. iPhone

Since I was using the iPhone as an exposure meter, I have a lot of comparisons between the two. Furthermore, the iPhone lens is equivalent to 28mm, which is the wide end of my normal zoom.

For one, I found that the iPhone has a pretty similar dynamic range to Velvia.

Velvia

iPhone (+0.4 stop)

In this case, I like the Velvia colors better.

The iPhone's auto white balance can be very different from the daylight-balanced Velvia:

Velvia

iPhone (-1.3 stop)

So which is better? I've seen it go both ways:

Velvia

iPhone

Above, I think the iPhone's auto white balance was fooled by the red leaf litter, giving the overall image an undesirable bluish tint.

Velvia

iPhone (+0.5 stop)

In this case, I think the iPhone did the right thing, and shooting daylight-balanced Velvia on an overcast day (which has a bluer color temperature) makes the yellow-orange leaves appear greenish.

Velvia

iPhone

I like the Velvia better here because of the better hue separation between the fence and the sand, but the warmness of the iPhone shot is nice, too.

Velvia

iPhone

The iPhone shot looks a lot better here, with an incredibly blue sky and a building that just pops. I don't know what happened. It's not the exposure level; maybe it's secret iPhone image enhancement?

Velvia vs. DSLR

In this case, DSLR = Nikon D90.

DSLR

Velvia

DSLR, Photoshopped to increase contrast

And that's pretty much it. I think post-processing can match Velvia's color saturation (I accidentally overdid it in this case), but the colors look funny when you push it like that. I think Velvia does a much better job capturing the dark red here.

The next example was taken during a cloudy day. Besides the bright sky, there isn't much dynamic range in the shot.

DSLR

DSLR, Photoshop auto-contrast

Velvia

iPhone

iPhone HDR

This is a pretty poor comparison photo. Velvia loses big time because of the white balance (blue clouds yield teal grass and greenish yellow leaves). The DSLR with auto-contrast looks pretty good, better than the iPhone. The iPhone HDR is pretty interesting. Otherwise not much to see here.

Velvia vs. print film

Purple flowers! For this test, print film = Fujicolor 100:

DSLR

Print film

iPhone

Velvia

I like the colors on the print film better than the DSLR. iPhone is pretty interesting: I think the blue is due to the iridescence of the flowers, and the iPhone's wide field of view and small absolute aperture make it especially apparent. Finally, the Velvia: the color on the screen is different from the color on the slide. I think it's a scan issue, because the slide color is a closer match to the other images. But the saturation, especially on the leaves, does look nice.

DSLR

Print film, low-quality scan

iPhone

Velvia

The close-ups are pretty revealing, though probably irrelevant for the vast majority of situations. Note that any compression artifacts you see are due to Blogspot/Picasa, not the original image.

The DSLR is incredibly sharp and free of any artifacts. The Velvia is also very nice, with slight grain on the shadowed leaves in the background. We can't really evaluate the print film except to say that Snapfish's scan is poor. The iPhone close-up reveals its "magic" at work: noise reduction (textured areas appear dead flat) and sharpening (ringing at the edges).

Conclusions

One final note: the scans are pretty different from the slides. I'm also a bit miffed because it looks like the scanned images were upsampled from a lower-resolution scan. So the scans aren't great. I'll be able to make stronger claims when I them all scanned at ScanCafe -- that should take out the scanning variability. For now, this is what I'll say:

• iPhone has poor detail resolution. Image may be 5 MP, but it can't actually resolve that, and it makes up with signal processing.
• Slide colors are awesome. Scan captures some of it, but the original slides make even the duller colors look vibrant.
• Contrast can be increased to make the DSLR look more like the Velvia, and I expect that more careful tweaking can match it better, at least with respect to luminosity (i.e., characteristic H-D curve). However, since recording color involves projecting the continuously-valued (infinite-dimensional) spectral irradiance onto a 3-dimensional (4, for some print film) space (e.g., RGB) through some spectral response function (which is different for different films and digital sensors), it will be impossible to get the color the same. And I think this difference is apparent to casual observation.