7 replies to this topic

[Paul Kay]

This last weekend we undertook some Seasearch (survey) dives off the North Anglesey coast. Despite calm conditions, the vis was between 1~2m and lower!

I was using a 50mm macro lens and had my camera set up to shoot both Raw and Jpeg images. The results from post processing the Raw images are astounding - whilst I'm sure the stretched tonality would give concerns if they were to be published as anything other than scientific records, the images yield tremendous dat which is simply not clearly evident in the jpeg images.

I'll try and post a couple of examples:

Attached Images

• 
• 

[JamesWood]

Good subject pgk

How scientists manipulate images depends a lot on what questions we are asking. If we want to count corals or DAPI stained bacteria, it is often a good idea to consistently adjust images so that the item of interest is easier to see.

On the other hand, when I’m specifically investigating something where color is a component, such as how well an octopus matches it’s background, I always shoot in RAW and uncompress the images in a lossless format like a tiff. I notice that in many cameras, the jpg’s that are actually manipulated to make the image look better. RAW was the only way to turn this off. I also include a grey card as a standard in every shot in case the lighting changes. These data images tend to be visually boring but they serve their purpose.

For pretty pictures to show the public and other scientists, we use our judgment to show what we perceived was there. Scientists have a LOT less leeway than artists, but we still have some wiggle room for pretty pictures. What we can’t ever do is misrepresent the truth of something. . . With words like “truth” and “perception”, we are heading more towards philosophy than science. . .

[herbko]

Good subject pgk

On the other hand, when I’m specifically investigating something where color is a component, such as how well an octopus matches it’s background, I always shoot in RAW and uncompress the images in a lossless format like a tiff.  I notice that in many cameras, the jpg’s that are actually manipulated to make the image look better.  RAW was the only way to turn this off.  I also include a grey card as a standard in every shot in case the lighting changes.  These data images tend to be visually boring but they serve their purpose.

<{POST_SNAPBACK}>

Color is a complicated topic. Shooting RAW does not guarantee accurate color reproduction. Outputs derived from RAW files can show as much color offsets as JPEG photos. You can easily see this first hand if you have more than one raw converter. Even using different camera profiles on the same raw converter can produce very different colors from the same raw file. A grey card definitely helps to get the right white balance, even with that it's still possible to have significant variations. A GretagMacBeth color chart would be better.

[JamesWood]

Color is indeed a complicated topic!!!

I'm using a grey card not to get the right white balance prior to the shot, but as a control to compare variation between a standard after the shot to control for exactly what you are saying.

For science shots, I don't create a different white balance each time, instead I pick a setting or kelvin temp and stick with it for consistancy.
Grey should have the same RGB values between shots, the grey I selected was in theory 120 120 120, equal parts of the three colors.

I've played with a MacBeth card but as far as I know, they don't make waterproof ones so it was a one time experiment. For qualatative teaching I prefered to use the colored squares but for quantatively showing the changes to the color with depth I used the grey squares.

Please see: http://is.dal.ca/~ce...andQuantity.pdf
for picts of what happens to a MacBeth card with depth in the clear waters of Bonaire, near noon, on a calm day.

[Paul Kay]

I too use a Macbeth card - but not underwater as they are too expensive!

On the subject of colour temperature, I see very varying numbers used to describe colour temperature from varying different cameras and suspect that different raw conversion software will also interpret the cameras settings differently too. It would be interesting to produce a much simplified and specific colour chart for underwater use - Depending on where you are (water tends to be green here in the UK) this could use a set of colour 'scales' based on the predominant colour (blue to green) as well as several other 'prime' colours - although not absolute, this would give at least an indication of more accurate colour adjustments which could be batch applied in photoshop for similar shots.

To get back to my original observations, I would say that the unadjusted, very dull and extremely low contrast shot would equate to what film would produce. The adjusted image makes it far easier (and sometimes even possible as opposed to impossible) to identify the creatures in shot and as such simply helps when undertaking surveys and the like.

[Kelpfish]

Shooting both raw and jpg, I have concluded that it is all in the lighting. If you have good light in a JPG you will get an image you like. Conversely, if you have bad light on a raw image, just because it is raw isn't a magic potion that will turn it into a usable image. It's true that you have more flexibility shooting raw, but nonetheless, garbage in garbage out, no matter if you are shooting JPG or raw. If your JPG's are hi-rez, well lit and good focus, and you have the same image in raw, you wont be able to tell the difference between the two.

Joe

[Paul Kay]

The point I am trying to make is that in the conditions we were surveying in (vis ~1m, near as dark as night) it doesn't matter how you light, the results are flat and low in contrast simply due to the severe diffusion due to large quantities of suspended matter. In fact my flash (a single unit as two would have accetuated backscatter even more) was almost directly overhead of the subject - at the ultimate backscatter minimising angle. In such cases the only way to adjust the image is by post processing, and raw shooting gives the greatest flexibility to do this.

Low contrast subject in + raw processing = informative, useful image

[JamesWood]

You both have made some very good points!

Often my scientific data recording images are far from my pretty picture images. While I can't tell the difference between the jpg and the tiff images extracted from raw files, my computer can as there are differences in color. For experiments on color (camouflage for example) this becomes very important. For other work, it often doesn't matter at all.

Post processing can help and act like a filter. And yes, one can only do so much, esp with a not so good shot to begin with.

Most of my science shots are garbage in terms of pretty pictures but do exactly what they were taken to do. It is just a totally different "audience".