This post is part of a series of posts about image quality. This post is going to be looking at Stitching Images together to potentially raise the quality of an image. You can follow the Image quality series here.
The idea behind super resolution came from astrophotography and the principle of stacking images. Stacking is when you place one image above another forming a stack. When an image is taken the sensor records the scene exactly the same (if the light is consistent) except for digital artefacts like noise, which are randomly created and so different in each image. This is why stacking an image has the ability to reduce noise. With every two images that are stacked you are in effect reducing the amount of noise by half in relation to its corresponding ISO. Simply, four ISO 3200 images stacked have a similar level of noise to that of a ISO 800 image. The drawback is that you need to have a scene with no movement, because movement will be seen when the stack is created. In astrophotography this process of stacking images enabled photographers to create the look of a longer exposed image, showing star trails in the sky, at the same time as reducing noise created from a longer exposure.
Photographing for Stacking
All of the images need to be taken from the same point with no movement in between, this means that you need a tripod. On a DSLR it would also be a good idea to lock your mirror up so as not to have any reverberation from the mirror moving. Your subject also needs to be still. If you are trying to recreate a faux long exposure you will need an intervalometer that will take a picture at the same exposure at regular intervals.
Stacking in Photoshop
In Photoshop go; File – Scripts – Stack and browse and choose your images. Alternatively you can open all the images and then select the station and choose the option Open Images.
Once you have your files stacked a tiny bit of math is needed to make sure the mean of all the layers is showing through. The equation is 100/n = Opacity %, meaning 100/ the layer number = the opacity for that layer. If you have four layers your opacity settings would be – Background 100% Layer 1 50% Layer 2 33% and Layer 3 25% Layer 4 20% – you have to treat the background layer as layer 1 in the formula.
For this post I took a photo of my cactus.
I took the image on my smart phone, with no tripod I set the phone to burst mode (taking 20 photos in a quick burst) and then chose the first four to place into a stack. When stacking the images, because I didn’t use a tripod, I chose the setting align images.
The difference in the level of noise is quite impressive and wondered what 20 layers would look like.
Original on the left, 4 Layers Stacked in the centre and 20 Layers stacked on the right.
With the 20 layer version the difference is slightly noticeable as it has slightly smoother tones but I feel that it isn’t as good as it could be, because when you get past 10 images you are dealing with decimals which can’t be set as an opacity setting. The reduction in noise has improved the image quality but still cannot be printed above its basic size.
This is where super resolution can come in.
Super Resolution
What super resolution does is enlarge the image and then using the different exposures with different pixels in each place smooth’s out any artefacts creating a larger image than can be made with a single exposure from the same sensor. This process is good when needing to enhance fine detail that or developing more accurate pixels for an image. As with stacking it does have the benefit of reducing noise as well.
For super resolution your process will be similar but with two differences. First you don’t need a tripod but you do need to take shots at a faster shutter speed than your focal length to keep the images sharp (eg 50mm lens your shutter speed can be no slower than 1/60 sec). This is because just like when I created images for stitching you need a slight overlap of the images, though unlike stitching you need a 99% overlap. The best way to do this is handheld as between each shot you will have moved slightly. This slight movement between frames will not be a problem as Photoshop will be able to realign the images.
Super Resolution In Photoshop
Screen Shot Adobe Camera Raw
Once your images are shot, you will need to open them in Adobe Camera Raw. Here make no adjustments, at the bottom click on the output options and set the resolution to the highest + setting (this setting is upsampling the image to its largest workable size by using a best match system to fill in the gaps of missing pixels as the image is enlarged). Load the files in a stack and use the same formula as when stacking images.
Due the new image file sizes, your final stacked file size will be considerably larger. The comparison between the original image and this will be astounding. My new super resolution image has a resolution of 240PPI and a print size of 19”x24”, making it a 26 megapixel image. Compared to the original with a resolution of 240ppi can only print at 10”x13”. Using super resolution has doubled the size of the image.
This is a comparison of a 4 Staked image on the left and the Super Resolution image on the right zoomed in at print size. This shows that the stacked image is of a lower quality at its smaller print size than the Super Resolution image.
In comparing the two processes Super resolution is the better method for improving the resolution and reducing the noise in the image. Although this process doesn’t fix any issues with the lens, exposure setting or composition. It also needs a bit of planning and forethought when it comes to capturing the image. But in a pinch knowing how to do this could mean a low quality image can be raised up.
Remember if you liked this post to; like, share and subscribe.
If you wish to get notifications when I post on my blog, you can follow me on Twitter@apertureF64, on Facebook.com/aperturesixtyfour or alternatively be emailed by subscribing below.
All images are the Copyright of Benjamin Rowe , ALL RIGHTS Reserved unless credited to another photographer.
For more information please read my Copyright Statement
Aperture64 
wow…you lost me from the very beginning…love the cactus though! 🙂
The idea is simple but maybe the process is confusing. Multiple images taken at the same time, on a low resolution camera, can be layered to create better image quality.
I am glad you like the cactus. It is my only plant i have not killed.
No, it is probably just me…I often understand things better by doing rather than reading. Love the cactus… 🙂
Is there some sort of an action or script that can be done for this that can get this performed on an automated one-click kind of thing?
thank you for sharing your knowledge!!
Cybele, no problems sharing my knowledge is something I like to do.
hy
i also tried your approach with some high iso night images, and I ‘confirm’ your findings. Mainly contrast is better.
Three questions tough:
– do you have a possible explonation why?
– why dont you use a smart filter mode? (no decimal problem 😉 )
– to compare enlarged vs. unenlarged method, what downsizing method do you use ?
cheers
darius
Hi Darius,
An explanation to why the contrast improved when stacking images is that the noise artifacts usually have lighter tone than the surrounding pixels creating a less contrasted image. Since noise is random, noise is not in the same place with evey image. Meaning that when you stack an image noise is in different places and start to cancel each other out creating more contrast. Also noise is more predominate in shadow and dark areas of an image which will exaggerate this contrast effect.
I do use smart filters and smart objects. I don’t tend to do stacking with every image as these steps are design for pushing the image quality beyond what the camera produces. For example when shooting at really high ISO, the Noise reduction filter won’t be enough but you can begin the process with stacked images and then apply a noise reduction filter.
Downsizing I use Photoshop’s Image – Resize; first changing to the desired resolution and then adjusting the pixel width on the longest side.
I hope that helps, if you have any more questions let me know.
for two years i’ve been struggling with this digital camera. it’s a decent camera. but regardless, the shots never seem crisp. i spent too much money on a new lens, and even with that, i still find myself sharpening in photoshop. is that normal? is it noise from shooting over 400 iso? i never had to think about this stuff before. and i don’t like doing so much work in post production. JT
All images need capture sharpening to counter the softening in digital capture. Good lenses will creat crisper images which is emphasised by using the len’s sharpest aperture. Noise on the other hand is something we have to deal with and basic noise reduction is also needed. If you shoot raw all of this can be done in raw conversion.
There is no need to follow the process above with all images just in those that are suffering the most eg when I have had to photograph in a church above 2000 iso.
Huh very interesting. Makes an impressive difference when you’ve planned for this with the necessary shots. Thanks for sharing.
If I am using my smartphone and know I want to print or display quite large I will use tge burst setting so I take advantage of this method. It is just something you have to think about hefore you press the trigger.
Makes sense. Thanks.
VERY good to know! Thanks!
Wow! I didn’t even know you could do this. You do a great job at explaining the process!
Yep, I do it quite often with pictures taken with my phone. It can also be automated which makes life even easier.
I have to apologize. My computer is having issues with responses on WordPress. The comment that was supposed to be on here posted on your other post I had commented on. Then it showed me that it hadn’t posted so I posted the same thing again. It now shows up twice over there. Ooops 🙂
“I have a question. If I only take one photo can I then make several copies on separate layers in Photoshop to achieve the same effect? I could maybe add a High Filter pass set at different strengths in relation to the percentages of each layer then put them on overlay for the blending mode?”
I don’t think it would have the same effect. The purpose of having multiple exposures is because Luminance noise and artifacts is random and will be different in each image. As you reduce the opacity you are blending images with different information ( ie noise) at different pixels creating a smoother image.
Using super resolution and then blending, as photoshop enlarges the picture it copies information form pixels and spreads it across a larger area. Having multiple images with different information really helps keep the tones smooth.
I hop that makes sense, if you have more queries just ask.
I know wordpress can have tech issues I added your comment to the beginning of my reply.
That does make sense 🙂 thank you for taking the time to explain the reason. I guess i should set my camera to continuous shooting to be able to try this…that’s easy enough to do 🙂
Continuous shooting would be good, give it a try and see what happens.
I will definitely do that. 🙂
Thank you for taking time to instruct us, Ben. So, we can do continuous shooting for the same spot, then to stack them?
yep that is exactly it, as long as the shutter speed is about 1/125 of a second or faster. This is because there will be less movement between frames. Once they are stacked they can all be aligned.
I’m with Heather above. I love the final super resolution image and understand maybe half of your description. I think I would probably have to be watching you do this, or do it myself as you are standing over me. But since you can’t be standing over me, I will have your post on my lap top and photoshop on my desk top and hope to duplicate your results.