Pseudo Long Exposure

Hi All

I was wondering if Hugin is capable of doing a pseudo long exposure, i.e. stacking multiple images of the same scene, let’s say 10 x 5 second images of a fast flowing river, to get a similar result to a single exposure of 50 seconds.

I have tried but I think that the anti-ghosting is preventing the effect that I want.

Any suggestions or perhaps, should I try other software.


You may try aligning images using align_image_stack, and then averaging them: How to average multiple images into one with ImageMagick? - Super User

1 Like

This is more in Siril’s wheelhouse than Hugin’s - although align_image_stack with hugin_stacker works reasonably well, I’ve found that for the use case of tripod-mounted synthetic-ND (e.g. no need to align), Siril is superior.

Getting a properly tagged DNG out of Siril is a bit of a PITA though (you need to save as FITS, convert to TIFF with imagemagick, rename to DNG, and tag the DNG with the appropriate color profile and CFA pattern metadata with exiftool), at some point I’m just going to write a Python script to handle this particular task with libraw and tifffile


Thank you, I’ll give the align and average a go.

Darktable has a lua_script called image_stack. It aligns images (if needed) with align_image_stack.exe and and then averaging (mean) them with magick.exe

Could you MarcoGiai-Coletti possibly send pictures so we can test if Hugin works with them. I have an idea that Hugin could work.

1 Like

@patdavid has written about this: Pat David: Faking an ND Filter for Long Exposure Photography

1 Like

Thanks – I was about to recommend that article, but was unable to find it: I didn’t go back far enough in time. @patdavid has been publishing articles since the beginning of time :slight_smile:.


I have done the imagmagick approach in the past, the issues I had with it were:

  1. IM only does internal calculations at int16 in most cases unless you recompile it specifically to do higher bitdepth internal math. At least 3-4 years ago, nearly all distros only packaged int16 support
  2. For large image sets, it would run out of memory

Siril at least uses higher bit depths for the internal pipe, but did have the issue that it saved to signed int16 when exporting to FITS two years ago, and required you to convert the entire sequence to FITS first too. These issues may have been fixed since then though.

(In my case, in addition to averaging out motion, I also wanted to average out noise to improve dynamic range).

I believe you can also use G’MIC to mean/median average a stack of images faster than I did in imagemagick (@David_Tschumperle can correct me here as he wrote the function for me ages ago).

I really should dust off my camera and actually start taking photos again (some blustery conditions here lately would have made some neat images of pilings in the surf at my beach). @MarcoGiai-Coletti please do post any experiments or results you get - would love to see them!

Thanks for all of the suggestions. The easiest and quickest option was doing it in GIMP with the G’MIC plug in. I tried it with a set of photos of rain on a lake but it didn’t smooth out as much as I’d hoped because I only averaged 10 shots. I’m going to find a fountain and try it with that.