Thursday, May 29, 2008

Would it be unethical to publish this image?

I'm a field-based structural geologist/metamorphic petrologist. I try to sort out when rocks were squashed or heated or buried or cooled, and use that timing to make arguments about what happened in the middle or lower crust. And that means that my evidence is often visual: field photos of folded dike, or thin section photos of foliations, or backscatter electron images of minerals frozen in the midst of an ancient chemical reaction.

Back in the day, I used film. I'm not a great photographer, but I learned to develop black and white film to try to get decent images of thin sections. (Then I laid a piece of mylar film over the photomicrograph, sketched the outlines of the minerals with a pen, scanned the drawing, and re-did the entire thing on a computer in Canvas. And published the two images side-by-side.) I also would take many exposures of everything I photographed, because I could never tell what would show up well in the developed film. And backscatter images... those I took using big pieces of polaroid film, and it was hard to guess just what combination of contrast and brightness would work.

Now, I don't do any of that. I've got a little digital camera that I take into the field, a digital camera on our polarizing microscope, and digital images of every mineral I even consider probing. But I may be sloppier as a photographer. A photo might be a bit too dark, or too bright. I leave off the tungsten filter on the microscope, for instance, and then take a set of adjacent pictures, knowing that I can easily splice them together in Photoshop.

When I'm done, my photomicrograph looks like this:



Is it ethical to publish this? It's a doctored image, and an article in the Chronicle of Higher Ed says that means that it shouldn't be in a journal.

It's all automatic doctoring, though - I used Photoshop's automated photomerge command to splice the photos together, merged the layers, and used the auto color adjustment to fix the yellow color from the tungsten light. In the olden days, I would have taped photos together into a mosaic, or adjusted the microscope magnification to get the image I wanted. I would have used a tungsten-balanced film (after taking an entire roll of yellow images) to avoid the yellow color. I would have taken multiple exposures to make sure the image was not too bright and not too dark.

And the photo isn't raw data. You, the reader, can't do all the things that I did to make sure I was identifying the minerals correctly. You can't spin the stage and watch the biotite change color. You can't cross the polars and see that the andalusite is grey, and the high-relief mineral (kyanite??) is red. You can't put in the gypsum plate and spin the stage around (though I know you want to). You've got to trust me that the mineral I say is garnet is really black under crossed polarizers.

But you can see some of the things that I want you to see. See the little yellow staurolites? See those white things beside them? Those are pressure shadows... and the garnet doesn't have them. You can see why I think the kyanite has partially replaced the andalusite - and you can tell me whether you've ever seen a texture like that before.

Is it doctored data, an intellectual lie? Or is it an attempt to show you what I see when I look at the rock?

12 comments:

Erik said...

I say as long as you state what was done to get that final image then you are fine.

C W Magee said...

Read the fine print of the doctoring guidelines- they generally refer to things like contrast-enhancing a specific part of the image, or artistically removing noise/background, or other specific instances of trying to make something look better than it ought to be.
And if you want to save yourself some time and effort, white balance your camera before you start photographing, and you don't need to adjust the pics later on.

Arr Cee said...

If your modified photo has the original alongside it, and you've stated that the modifications are show what is actually there (as opposed to what the camera has recorded), then I can't see any problem. That said, I know nothing about the world of scientific publishing.

p.s. I loooove andalusite!

p.p.s. Sorry, I inverted a sentence for some reason.

Silver Fox said...

If the guidelines are actually talking about increasing contrast, or changing the brightness level - on even a specific part of an image - then I think they are being a bit picky. They have forgotten that anyone who did or does their own darkroom work with film photographs, or has sent them to a professional rather than to the corner drug, has probably done the same things to black&white and color film photos. And that if those things weren't done, the photos may not have been very good. Pros do it all the time.

It's fine to say to just set your camera "better" or in some different way, or to always use the "right" filter, but very few images I've seen taken by a digital camera couldn't stand some basic "enhancement," especially in the contrast department.

I think 1) check with the specific journal, 2) publish changed images side-by-side if necessary, and 3) doctoring minerals (or field photos) in geology for the purposes of making them easier to see and understand might not be (shouldn't be?) the same as changing photos of cell structures in biology.

I've taken lots of close-ups of minerals in rock specimens and through a scope, and have almost never been able to white-balance a camera completely for close-up or through-the-scope lighting. I'll do the best white balancing possible, and then still have to change the color some later. I'd rather change it and be allowed to change it in an editing program than to have off-color images published or sent to colleagues (can't spell that word). Color balance of film photos, if you print your own color prints in a darkroom, almost always has to be adjusted. Why any different with digital cameras?

Remember - almost all film photos used to be "doctored" or enhanced prior to publication - if not by the photographer (scientist in the darkroom) or the printing photographer (also in the darkroom), then by the journal itself.

I think the only question should be whether the scientist is trying to deceive or not - something that can only be determined later.

Kim said...

If I selected all the biotites by color and changed them to a nice violet-blue and claimed to have coexisting andalusite and glaucophane... now that would be fraud. (I know it's possible, but I'm not good enough with Photoshop to do it.)

The automated photomerge changes the color and brightness around the edges of the individual photos to make them match better (and distorts the images, as well, if necessary). I don't think the illusion of a single image is possible otherwise - when I've taped mosaics together by hand, I've found that the lighting generally isn't consistent across an entire photomicrograph, and that there may be some distortion of the image around the edges, too. (I've looked at this thin section since making the mosaic, though, and it looks as though Photoshop did a good job of making the image resemble what I see in the microscope.)

Anonymous said...

My personal opinion is that generally photo-capturing devices do not do the best of jobs at capturing what we actually see. I think there is responsibility to allow other researchers a look through our eyes to see what we see. If this means using a particular filter, then use that particular filter and tell communicate that choice. However, there are commonly accepted techniques in image analysis that are field-dependent, and you should not have to explain all of those all of the time.

Anonymous said...

I agree with the comments above ... I don't think researchers need to be worried about including some of the more standard photoshop techniques into their "methods" (e.g., levels, shadow/highlight, etc.). But, if one starts really changing colors and such, especially when those colors are representative of something discussed in paper, then I'd like to see some mention of what was done to the photo in the methods.

I remember listening to a ScienceFriday last year about how digital photo experts are able to go into a photo and "back out" the stuff that's been done to it.

Anonymous said...

I agree with the comments above ... I don't think researchers need to be worried about including some of the more standard photoshop techniques into their "methods" (e.g., levels, shadow/highlight, etc.). But, if one starts really changing colors and such, especially when those colors are representative of something discussed in paper, then I'd like to see some mention of what was done to the photo in the methods.

I remember listening to a ScienceFriday last year about how digital photo experts are able to go into a photo and "back out" the stuff that's been done to it.

C W Magee said...

Try looking up a planetary science paper, or something similar that you know is a prettified mosaic, and see how they describe their image processing.

Elli said...

I hadn't even thought of this issue... but my question is this: where is the photomicrograph from???

Kim said...

Elli - this is from Gallup Mills. Gile Mountain Formation, just west of the Monroe Fault. That streambed outcrop in the Moose River.

I've just probed it, and need to analyze the data. (Are you sure you don't want to work on the textures? After your PhD is defended and you've got papers in press from it? The outcrop is really accessible...)

Elli said...

I'd love to tackle it! But first, as stated, I need to get some papers submitted :)