Any benefit using D-log for photos?

[Tune]

Hi!

As I understand it, using D-log when you shoot video is great (if you colour grade it afterwards).
But is there any benefit in using D-log for shooting RAW photos as well ?

Best
/Mikael

 

[dslater]

Pretty sure it has no effect on raw images whatsoever.

 

[kevinhanna]

Edit: Arg. I just realized @dslater is right. D-log is used in video since Raw often isn't an option and if it is, the file sizes are so big that it creates other problems. The invisible data in the shadows is stripped away as part of the compression, D-log minimizes how much ends up getting stripped during the compression. Raw isn't compressed so doesn't strip anything away. So just ignore everything under this.

I'd say no. The benefits of using D-log in video don't really translate to photography. With video the light can change from the beginning of the shot to the end, so using D-log allows you to minimize both overexposing highlights and underexposing shadows, it brings everything to the middle and minimizes lost data (blown out highlights have no data to recover, it's pure white, and the the same is true for shadows become pure black). When you're taking photographs, you can adjust the exposure to minimize data loss for one (usually highlights) and then you're maximizing the amount of data you capture in the photograph and you'll be able to recover a lot in software. For example, if you adjust your exposure so that the clouds in the sky are only slightly blown out take the shot, you'll end up with a lot of black shadows, but there will still be a lot of detail in those shadows that you can pull out using your software of choice and you end up with a lot more dynamic range in the edited shot than you would using d-log (which minimizes having too much of one or the other but higher sum total of both). If you tried to do this in video and happen to pan towards the sun, you would end up with a blown out sky, and you would not be able to recover any detail in that sky, which is why d-log aims desaturate towards the midtones.

 

[Oyibo VFX]

I have some photos that I took after filming in D-Log and forgot to put it back. Did some colorgrading and actually, it came out very nice. Would I do it again..? Nope. Would I be angry if I had to correct again, also no.

 

[kevinhanna]

I have some photos that I took after filming in D-Log and forgot to put it back. Did some colorgrading and actually, it came out very nice. Would I do it again..? Nope. Would I be angry if I had to correct again, also no.

Yeah, good point. I've taken some shots that were not well exposed that would have been salvageable had I used D-log. I guess if you're goal is to get the absolute best shot, the D-log is not good, if your goal is to avoid taking unusable shots then D-log is great.

 

[Oyibo VFX]

Yeah, good point. I've taken some shots that were not well exposed that would have been salvageable had I used D-log. I guess if you're goal is to get the absolute best shot, the D-log is not good, if your goal is to avoid taking unusable shots then D-log is great.

I wouldn't say that... Also in photography, the same rules apply. A good photo is a good photo. While a bad one remains bad. Including out of focus

 

[Tune]

Wow, that was some quick replies =) Thanks!

 

[Oyibo VFX]

Wow, that was some quick replies =) Thanks!

Quick pilots here

 

[geigy]

Yes, the concept of dlog does translate to still photos, because you will have more headroom for adjusting color and contrast. But only if you are going to jpeg. You are probably better off shooting in RAW which also preserves a maximum amount of data.

 

[Droning on and on...]

What may not be obvious is this is really a discussion about Dynamic Range and quantization error.

The dual-ISO sensor gives more stops at the high end, so use 800 and above for bright scenes to reduce/eliminate washout. using below 800 switches to the stronger analog amplifier in the sensor reducing noise in low light, extending stops at the dark end of the sensor's range.

RAW doesn't help preserve detail lost in the ADC due to quantization error – analog levels between discrete digital levels that in fact represent detail in the shadows and dark areas, but is lost when they convert to the same digital value.

A log transform to the analog signal before ADC can preserve a lot of detail, but must be reversed in post, where there are more bits (levels) to work with.

So what you really want is dlog raw.

 

[retiredNH]

What may not be obvious is this is really a discussion about Dynamic Range and quantization error.

The dual-ISO sensor gives more stops at the high end, so use 800 and above for bright scenes to reduce/eliminate washout. using below 800 switches to the stronger analog amplifier in the sensor reducing noise in low light, extending stops at the dark end of the sensor's range.

RAW doesn't help preserve detail lost in the ADC due to quantization error – analog levels between discrete digital levels that in fact represent detail in the shadows and dark areas, but is lost when they convert to the same digital value.

A log transform to the analog signal before ADC can preserve a lot of detail, but must be reversed in post, where there are more bits (levels) to work with.

So what you really want is dlog raw.

Uh, no. Bits are bits, whether logged first or not. When decoded, you only get as many bits in output as you had on input. Because video is ALWAYS compressed (as was previously stated) a log transform, with the right parameters, can have a positive effect, depending on the result sought by the videographer. Because still raw is NOT compressed, you don't get the same benefit.
Worse, because still photographers tend to use different photographic approaches, both to source and to output (post-processing), a log transformed image could result in unexpected, undesired banding.

So, yes, this is about compression - that that takes place in video. It doesn't happen with still raw.