4

I attended a Kelby One workshop where he demonstrated the median combine method to blend multiple images and remove tourists, cars, etc. (i.e., anything that moved between shots). However, that particular tool is only available in CS Extended and CC.

Can this process be emulated in CS6 (NOT Extended)? If not, can it be done in GIMP?

Thanks!

2
  • Anything can be done in GIMP! ;-) Your comfort level may be different though than with Photoshop! :-D
    – TFuto
    Apr 21 '16 at 17:55
  • Believe me, I'm a big fan of open source. It's just that darned learning curve that goes along with so much of it. :-) Apr 23 '16 at 2:21
5

Yes, I have done it in CS5 and CS6 with the following:

  • File > Scripts > Load Files into Stack
  • Select all layers and use Edit > Auto Align to align them
  • Layer > Smart Objects > Convert to Smart Object
  • Layer > Smart Objects > Stack Mode and choose Median
0
4

There's a tool called imagej (or its newer equivalent fiji) designed for use on scientific images. It does this out of the box (assuming the images are aligned). You'd probably need to save as .tiff.

It's free, open source and cross platform. Very old versions are sometimes still found which treat each colour channel separately but I've used newer versions that don't do that.

3
  • Indeed and it's quite easy to write your own plugins for ImageJ, I've written one specifically to do what Oliver wrote in his answer, where you create your masks to remove the unwanted people, blur the masks a bit and then the take the weighted average of all the pictures, where the weights are given by the masks. Apr 21 '16 at 19:21
  • @CountIbis now that's interesting. I've only used it for microscope images myself and didn't try writing anything for imagej (the analysis used python with pyfits).
    – Chris H
    Apr 21 '16 at 19:25
  • I have used AstroImageJ for astrophotography. Never thought to try ImageJ for this. Thanks! Apr 23 '16 at 2:22
3

A (better ?) alternative method using Photoshop CS6 exists :

  1. Import your photos using basics script : file -> Scripts -> Load files into Stack
  2. Align : Select all the layers and Edit -> Auto-align layers
  3. Add a Layer Mask to all the layers (all your images)
  4. Select a brush and paint white/black on the mask to hide/reveal what is behind your current layer

The order of your picture is important ! If the image part you want to keep is on the last image (behind all the others), you have to make the corresponding part of each of the layer mask on top of it black to reveal the good part. Use the visibility icon may help.

I think that this method works better than the mean filter, unlesss you have a looot of noise to remove. Indeed, if you have moving things on your photos, like grass, herbs, trees, a mean filter will yield unnatural effects whereas this method will keep coherent most of the image part.

If you want an illustrated example, take a look here : http://www.photoshopessentials.com/photo-editing/how-to-auto-align-and-composite-images-in-photoshop/

10
  • This is much better than median, when you expect that every image has the same exposure, there was no shake, etc.
    – TFuto
    Apr 21 '16 at 17:42
  • 1
    Well, technically a median filter is a kind of averaging filter, and suffers from the same problem: an averaging filter is a low-pass filter, so it will remove some noise but also will remove some high-frequency detail, so your contrast, sharpness suffers. This is even true if you don't have alignment and different lighting issues, but have some ISO or other noise on each image... Low-pass filtering the noise ends up removing sharpness as well.
    – TFuto
    Apr 21 '16 at 17:52
  • 1
    Yes, it is always the noise/contrast/sharpness battle. They made a good example of it here : cambridgeincolour.com/tutorials/image-averaging-noise.htm
    – Olivier
    Apr 21 '16 at 17:56
  • 1
    Averaging usually works well on the very same image. E.g. if you scan an image 100 times, then after the averaging the (random) noise variance will be 1/sqrt(100), so 1/10. This is the way to reduce random noise coming from the scanning process. (It does not handle systematic noises, that happen at the very same places.)
    – TFuto
    Apr 21 '16 at 18:08
  • 1
    @TFuto averaging within an image is a low pass filter, averaging between images is not. It can actually increase sharpness by increasing the signal to noise. What you lose its in the time domain, and the problem there is that almost everything moves (e.g. trees in the tutorial posted above). If the subject and the camera are all hard edges and rock solid, this is a very powerful technique..
    – Chris H
    Apr 21 '16 at 18:47
0

It's possible in GIMP as well:

  • File > Open As Layers
  • Filters > G'MIC
  • Set Input Layers to "All" in the dropdown on the left
  • Layers > Align Layers, Apply
  • Layers > Blend [median], OK
0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.