Stabilizer or faster lens?

[String]

Regarding the colors at depths: red is there waiting to be extracted with proper filter and processing.

No it's not. You can't cheat physics. Red and other longer wavelength light gets attenuated very rapidly by a water column. In other words its reduced and eliminated. The ratio of reds and oranges to the blues and greens dimishes to the point theres none left or at best a tiny bit left which you then amplify creating lots of noise.

You also seem unaware as to what a filter does. It doesn't "extract" red or put red back in. All a filter does is remove blues and greens to reduce their influence to try to correct that balance. Once you get below about 10-15m balance simply isn't possible as theres no red left so all it does is cut down on light thereby needing an even slower shutter speed or higher iso.

You can see the problem very easily for yourself by just looking at an RGB histogram of photos of various depth. You can see the red channel get weaker very rapidly before disappearing completely in the mid teens depth. It's gone, not there to be extracted, gone.

Ive glanced at a few of the photos youve posted and its a case of a lens that isn't wide enough so you're too far away combined with too deep for ambient light giving the horrific magenta tinge and added noise as you've desperately amplified the weak and non existent signal from the red channel. You've pushed a really substandard lens setup too far and gone too deep with a filter which accomplishes nothing. U/W photography is all about knowing the limits of the setup and operating within those to produce acceptable photos.

 

[olegk]

But then, how do you explain that, for example, the picture at DSC03665.JPG, taken at ~28.7m does have enough reds?
The red channel is weak, but not inexistent.
The filter reduces greens and blues, which are the dominant colors there. As a result the exposure is enlarged, allowing the sensor to capture more reds. When I tried FLD filter at these depths, I had to reduce black point in the RAW converter to prevent red channel from falling to zero. But FLB did much better.

 

[CT Sean]

Regarding the processing alternative, it's obvious that water bulk looks better in Mr. CT Sean's approach. But the color of the diver in my "original" is closer to the reality. That's the restriction of the method - you cannot have both. E.g. with ambient-light UW, it's better to have less background. On land this would mean longer lens.

I have to admit I find it surprising that the diver in the photo with the same purple color cast as the rest of the image is actually a proper representation of reality. Either way, sometimes in a photograph, reality has to suffer for the sake of aesthetics.

Post the RAW file somewhere and we'll see what we an really do with the image.

 

[tursiops]

The OP already knows it all, so why are you guys bothering? Let him live his fantasy.

 

[String]

But then, how do you explain that, for example, the picture at DSC03665.JPG, taken at ~28.7m does have enough reds?

It doesn't - its not a balanced exposure. You've got an underexposed foreground and hugely over exposed background (you're not claiming it really looked bright white at that depth?) and whatever tiny amounts of data left in the red areas has been boosted (and a resulting increase in noise). Simple physics dictates you've lost a lot of the red, orange and yellows at that depth and you cant put back what the water has eaten. Artificially boosting a very low residual signal is not the same thing.

The filter reduces greens and blues, which are the dominant colors there. As a result the exposure is enlarged,[/quote]

Nope

allowing the sensor to capture more reds.

Nope. You need to read the basics of what RAW is and how sensors record. It isn't a case of the blues and greens fighting for space over the reds. We're talking different photosensors for each one operating completely independently. What a raw file does is gives you access to that complete unedited data off the sensor. Simply removing blue and green isn't going to magically help a sensor record more red, especially if the laws of physics have already eaten it!

There's a reason people spend a fortune on lights and not filters and thats because filters don't work in any way near the way you describe - they can help adjust a colour balance in the shallows where plenty of red/orange wavelengths are still present. But all they do is reduce light, nothing more complicated than that. Arguably, shooting RAW a filter really isn't that useful anyway.

---------- Post added December 15th, 2014 at 05:49 AM ----------

The OP already knows it all, so why are you guys bothering? Let him live his fantasy.

This is true although replies are useful for the benefit of other people who may find this thread in a search and come out with completely incorrect information.

 

[olegk]

Could anybody please explain what "OP" means? If it's applied to myself, should I know it?

Then, while the statement "Simply removing blue and green isn't going to magically help a sensor record more red, especially if the laws of physics have already eaten it!" is generally correct, it's not very relevant.

The filter reduces green and blue, thus allowing to raise exposure without saturating green and blue channels; as a result the otherwise very weak red channel is amplified. That's physics too. Yes, there red channel is noisy, but I can live with it.

What I'm trying is to do as good as I can with reasonable-price + reasonable-weight + reasonable-time-spent-on-processing method. Hope this sounds, well, reasonable. I know there are restrictions, and some pictures won't ever look realistic, but this isn't always required.

There are restrictions with UW flash photography too, not even mentioning that you are bringing to the UW environment strong light that doesn't belong there, and I cannot be sure that some organisms don't get hurt by it.

 

[Storker]

Could anybody please explain what "OP" means?

OP = Original Poster, alternatively Original Post

Then, while the statement "Simply removing blue and green isn't going to magically help a sensor record more red, especially if the laws of physics have already eaten it!" is generally correct, it's not very relevant.

The filter reduces green and blue, thus allowing to raise exposure without saturating green and blue channels; as a result the otherwise very weak red channel is amplified. That's physics too. Yes, there red channel is noisy, but I can live with it.

You're partially right. However, the red wavelengths are attenuated very quickly underwater, that's inevitable, and below 15-20m there's hardly anything left at all. Just look at the histogram, and you'll see that. You can't recover what's not there. That means that even with a red filter that reduces the intensity of the blue and green channel and thus enables you to increase exposure without saturating those channels, you'll struggle to get good reds and oranges at depth using only ambient light.

Now, ambient light photography is a very interesting sub-genre, and also a fascinating task to try to tackle, but you will never get topside color balance. If I were you, I'd rather accept a blueish/greenish tint to the picture - alternatively, convert to B&W - than try to mimic in post the colors you get from using a light or a flash. That's physically impossible. If you want the reds and oranges to 'pop', you will need some type of artificial light.

Some ambient light photos I've taken:



A torch may be a low-cost alternative to a strobe, albeit with less control of the lighting.

Using a torch:

 

[tursiops]

This is true although replies are useful for the benefit of other people who may find this thread in a search and come out with completely incorrect information.

Excellent point.

 

[CT Sean]

I was under the impression that when shooting RAW, a red filter isn't accomplishing anything anyway - maybe making AWB closer in-camera and requiring less WB correction in post but that's it. The key being images aren't actually gaining anything from an image perspective by using one (when shot in RAW).

 

[Storker]

I was under the impression that when shooting RAW, a red filter isn't accomplishing anything anyway

IMO you're wrong here. If you're at intermediate depth, a red filter will reduce the blue and green channels, require more exposure and make the red channel a bit stronger since you expose more without blowing out the B and G channels. That means you can recover whatever's left of red without suffering that much noise in the R channel.

If you're so deep that there's no red light left, you're only increasing the shutter speed...


--
Sent from my Android phone
Typos are a feature, not a bug