Combine multiple shots to simulate long exposure?

[Guillermo Luijk]

If you really want to achieve the same as a long exposure you need to:

1. Make the delay times between shots negligible compared to the duration of each shot (if there are bright moving objects such as lights, some gaps can still be visible in the resulting image such as those seen on mutiexposure star trails).
2. Average the images with the following weights (upper to lower layer):
For 2 shots: 50%, 100%
For 3 shots: 33%, 50%, 100%
For 4 shots: 25%, 33%, 50%, 100%
and so forth
3. Use a linear (gamma=1) setting in Photoshop. No one has mentioned this, but as long as you use any gamma profile the result will not be a reliable simulation of a single exposure (remember a sensor is a linear photon counter).

If you accomplish with those 3 rules you'll achieve the same visual effect as a single longer exposure, but it will have a much higher noise quality since averaging will improve SNR.

Another drawback is that sharpness can be reduced because of minimal misalignment if you cannot lock the mirror up during all shooting time.

Regards

[Tony Jay]

Thank you Guillermo.

Although I did not start this thread I have found it most instructive.
In the next few days I will test your suggestions.

Regards

Tony Jay

[Christoph C. Feldhaim]

3. Use a linear (gamma=1) setting in Photoshop. No one has mentioned this, but as long as you use any gamma profile the result will not be a reliable simulation of a single exposure (remember a sensor is a linear photon counter).

WOW - I just had overlooked that.
Should have known additive operations are not allowed at non 1.0 gamma....

* Christoph C. Feldhaim slaps forehead

[BobFisher]

The four pictures look very similar to each other.  I'd expect the combined pictures to look much smoother than the original images, as the scenario I'm trying to deal with is that it's too bright to take a single exposure that's long enough to sufficiently blur the image.  By combining a number of pictures, each with a relatively short exposure (so the water is not smooth in any one picture) you end up with a smooth image.

I may well be confused about exactly what you are doing

They do look similar at such small size.  But if you do the same things on larger images, you'll see the differences, which are in the water areas.

3. Use a linear (gamma=1) setting in Photoshop. No one has mentioned this, but as long as you use any gamma profile the result will not be a reliable simulation of a single exposure (remember a sensor is a linear photon counter).

This is achieved how GL?

[wolfnowl]

This might be of use to you: http://www.wolfnowl.com/2010/12/photographing-moving-water/

Mike.

[Guillermo Luijk]

WOW - I just had overlooked that.
Should have known additive operations are not allowed at non 1.0 gamma....

* Christoph C. Feldhaim slaps forehead

Additive operations are 100% allowed at non 1.0 gamma, but they will not exactly mimic what you get in a single exposure (I did a comparision one day but cannot find it now, will repeat it again tonight just for curious). I was just being strict here though since the results between adding images using 1.0 gamma profiles and any other arbitrary gamma will not differ much; in fact you may even prefer those obtained using gamma profiles. In addition to that the 1.0 gamma images must have been neutrally developed (not contrast, bright or saturation adjustments yet), then added, then processed.

This is achieved how GL?

Just need to convert images to linear versions of your profile (sRGB, Adobe RGB,...). You can build a linear version of any profile straight in PS. I explained how in the DCRAW Tutorial, in the 'COLOUR PROFILE' section near the end. But as I said, this is just to be strict, I don't think the mess is worth.

Regards

[BobFisher]

Just need to convert images to linear versions of your profile (sRGB, Adobe RGB,...). You can build a linear version of any profile straight in PS. I explained how in the DCRAW Tutorial, in the 'COLOUR PROFILE' section near the end. But as I said, this is just to be strict, I don't think the mess is worth.

Regards

Converting the images to the LinearRIMM profile available from Color.org would also work, no?  That assumes that the RAW image is tagged with the ProPhoto space to begin with.

[milt]

Unless you are absolutely fixated on Photoshop, there is an easier way to do this.   PhotoAcute will smooth out the water, do ghost removal on the static parts, improve noise over the static parts, and as a free bonus, quadruple the number of pixels.  I have found it best to feed PhotoAcute lots of images (I see improvements up to around 15-20).  I have never tried hand-held shots.

--Milt--