Fibonacci Exposure Bracketing

[garry23]

Some may have seen this, but in case you haven't, you may interested in this article. It might stimulate the ML gurus ;-o)

http://www1.cs.columbia.edu/CAVE/publications/pdfs/Gupta_ICCV13b.pdf

[budafilms]

Guau! A1ex is the men to see this.

[Audionut]

Hardware Prototype:
For generalized registration, a sensor that allows exposure bracketing with a negligible inter-frame time-gap is required. Although most current cameras support exposure bracketing, there is often a large inter-frame gap (50−200ms). We implemented our techniques on a machine vision camera which is triggered externally using a micro-controller based circuit.

Have a look at dual_iso.  It doesn't need hardware prototypes 

[garry23]

Fully aware of dual ISO etc and a user.

Just thought some may enjoy reading this article.

I'm not 'requesting' anything!

Cheers

[eduperez]

Fully aware of dual ISO etc and a user.

Just thought some may enjoy reading this article.

I'm not 'requesting' anything!

Cheers

Well, I did enjoy reading the article (just skimmed through it ); thanks.

That paper deals with image registration (alignment) and bracketing strategies. The first part is not very interesting to me, as it concerns mostly to the post-processing phase. But the part about bracketing strategies could be relevant.

They have studied the best approach to capture a HDR sequence of images under a time constraint: how many exposures should be taken, and how long should they be, when there is a maximum time allowed to capture all the sequence. Under normal conditions, the traditional burst of 3 to 9 exposures, with 1EV separation between exposures, can be enough; but for low-light situations, where longer exposures are required, the total time needed to capture a scene can be a problem.

Instead of using a 1EV separation between exposures, they propose using a Fibonacci sequence; for example, instead of 1s, 2s, 4s, 8s (15s total), the propose using 1s, 2s, 3s, 6s (12s total).

[Audionut]

Are your midtones clean?

1s, 8s.  HDR captured.

[engardeknave]

1s, 8s.  HDR captured.

Orly? And what would be the process to arrive at a satisfactory result with just these frames?

[Audionut]

They are 4 EV apart, and will be satisfactory for a significant amount of HDR captures.

If the reader needs specifics, such as median noise reduction, 20+ EV or DR, or any other number of specific situations, then one would assume that the reader would use the grey matter in between the ears, and research the specifics of their situation!

I have no problems suggesting that captures of only 1 EV apart, are simply wasted shutter actuation's, for all but a limited number of users.
Most people would use 1 EV brackets, simply because they don't know any better, and that's what someone on the internet said to use!

If you need specific bracketing for your purposes engardeknave, then use your own workflow specific to your purposes.

Feel free to explain to me the benefits of using 4 bracketed shots 1 EV apart, as opposed to only 2 bracketed shots 4 EV apart.

edit:  In the advanced bracketing feature request, where such things as 0.3 EV bracketing were being requested, and were basically ignored, since it would only benefit an extremely small subset of users, dmilligan went on to try and explain the benefits of tight EV spacing. 

Funny thing is, not 1 single person joined in on the conversation.  I guess it's extremely easier to just say you need something, then to try and explain why you need something.

[engardeknave]

will be satisfactory for a significant amount of HDR captures

specifics, such as median noise reduction, 20+ EV or DR, or any other number of specific situations

Ok, if we're in agreement that "1s, 8s.  HDR captured [period]" is completely wrong in many if not most situations, then I guess the matter is settled.

Most people would use 1 EV brackets, simply because they don't know any better ... ... ...

That's not what you said or even related to it. Come on.

[Audionut]

Ok, if we're in agreement that "1s, 8s.  HDR captured [period]" is completely wrong in many if not most situations, then I guess the matter is settled.

Did you even bother to take the post above mine into context?  Or should I have quoted it to make it easier for you.

That's not what you said or even related to it. Come on.

It's all related to the discussion about how many frames are needed.  But you only seem interested in taking my general statements, and having an argument over specifics.

It's impossible to have a discussion about topics, in a very specific manner, considering all use cases.  To do so would require a scientific journal.
Remember that part I mentioned above, about the reader using their grey matter!

[engardeknave]

how many frames are needed

1s, 8s

I've done the math. I'm getting two from you. Are my calculations off?

[Audionut]

Still not taking the post above mine into context hey?  Is that a little to complicated for you?  I can change my post with sufficient quotes if that will help your maths.