Better Than HDR? Try Image Averaging for Better Results

[@stevep8553]

Thank you. Love to see bright young people being so confident in what they do. A smart young man - good luck.

[@dougsturgess2651]

Very interesting. Something I've learned about taking photos for exposure blending is for 99% of circumstances, as long as you have all the shadows & highlights captured in the histogram, just 2 images will be better than multiple. If the histogram contains all the information in shadows & highlights, the HDR will process great. The only time when you may need a 3rd is if you have such high contrast, a 3rd underexposed is needed. Try it & see. I used to take anywhere from 5-7 bracketed images for HDR. I get better results with just 2 most of the time. I've run into the artifacts with things moving before & didn't know how to handle it. Look forward to trying averaging. You're a great instructor. Glad I ran across your site.

[@biggsy..215]

Thank you for your input i will defiantly try that.👌

[@Powerstroke98]

I very rarely take more than three, but yes, two often is all that's needed, if your settings are correct, via the histogram.

[@britishrose9417]

How did I not know this earlier! The photos look far more natural than HDR. Thank you so much!

[@christopherkemsley4758]

First off, I really appreciate that you got right to the subject of the video without 5-10 minutes of preamble - thank you! Secondly, this is really good to know. I've only recently gotten into HDR photography and haven't yet had (or, at least, noticed) those artifacts, but now I know what to look for! Do you know of good software to batch-average photos? You made a comment about workflows ... editing sets of 2/4/8 images at a time, one by one, in Photoshop is not a particularly-reasonable way to handle a photoshoot where I take tons of photos...

[@PhotographyLifeChannel]

I think for batches like that, I’d still do traditional HDR! Although you could create a Photoshop action to speed up the process, image averaging takes a bit more time than classic HDR. No issue for a few stacks, but it could add up if you have several dozen or hundred.

[@christopherkemsley4758]

@@PhotographyLifeChannel I'm astonished at how well this worked. The moon looked pretty neat last night (though I just barely missed the clouds surrounding it) so I decided to try an HDR set and also use this technique. My HDRs didn't come out well (the time it took to bracket was long enough for the moon to move significantly) but I was able to get 16 photos to average together and ... wow! There are many places in the darkness where the original photo just faded into oblivion but the averaged photo using your method still has very nice details. Thanks for the video!

[@tonyjin7194]

Great video! I personally tend to swear by median when there are moving people, since you can make them disappear completely rather than just reducing their opacity to 25%. When there is moving foliage, I tend to use a selective blending technique where only one frame is used in moving portions (so long as the noise is manageable).

[@Namaste..]

Very clear explanation. Always find your videos helpful. Many thanks!

[@PhotographyLifeChannel]

Of course! Glad you liked it 😄

[@shengyetang7220]

Very interesting Spencer. I just did some test with my 5D mark III. I took 32 under exposed image of a high dynamic range sceene, then stacked them, and then open ACR in photoshop. When i revcovered it with exposure slider or shadow slider, the result was much better than single shot, however there were still many noise and stripes in the shadow. 1:Clearly, with my 5D mark III, Image Averaging is not exceptable at this kind of high dynamic range sceene. 2:Exposure blending is the best option at most situation, if anything moving, then we can choose Image Averaging. 3:Does my test result concerned to ISO Invariance?

[@StephenJWalter]

Great video and a new idea for me to mull over. Came across this video while looking for information on HDR, which I’m just about to get into. I’ll certainly add averaging as an extra technique to play with.

[@Tony-Elliott]

Excellent content again Spencer

[@PhotographyLifeChannel]

I’m glad you liked it, thanks, Anthony!

[@NataliaSkorokhod]

I didn't have an internet connection for the past few days, so glad to finally be able to watch this! This technique is really cool, I actually didn't know about it. I used to work a lot on manually fixing HDR artifacts :D

[@PhotographyLifeChannel]

Same here! It's really annoying to fix them manually. I'm a bit surprised that no one really talks about this image averaging technique instead. Not that it's completely flawless, but it really does work better 99% of the time when something in the photo is moving.

[@vikassaraf3122]

Your content is gold! Just beat the KZbin algorithm and u are gonna groww like crazy!!

[@PhotographyLifeChannel]

That’s the goal! It seems to put my videos reasonably high on Search but not much on Recommended. Maybe I need titles like, “HDR Is DEAD! Here’s the SECRET Trick To Use Instead!!”

[@NataliaSkorokhod]

@@PhotographyLifeChannel "The Photomatix Pro creators hate this guy!" Also don't forget to add the "HDR is DEAD!" part in huge colorful font to the thumbnail :D

[@Dunja0712]

You also need a huge photo of your confused face and lots of question and exclamation marks. :D All joke aside, YT algorithm is a bi*ch, but your content is great so I'm sure you'll be huge!

[@PhotographyLifeChannel]

@@Dunja0712 Thank you, Dunja! Definitely, and the more emojis in the thumbnail, the better.

[@quincylee2276]

hmmm...I'll give this a try for lunar photography and see how it compares to the HDR process. Thanks for the video!

[@peterdesmidt8742]

Great job! I'm looking forward to giving this a try.

[@thomastuorto9929]

Thanks for the vid. I don't have PS so didn't know about this. You don't think manually Exposure Blending multiple stacked layered photos is any good. More work but most likely worth it from what I have seen online. Thanks for any replies.

[@luciengrillet]

Nice tip ! I’ll definitely try it when shooting near the ocean during golden hour.

[@christophermason7735]

Very useful. Thanks. I have a blue hour HDR I’m really struggling with some weird purple ghosting in the trees. I’ve loads of shots that are not over exposed so I’m going to try this. I have paintshop pro ultimate and they have a photo blend app built in for removing moving objects. So I am thinking I will take 8 frames remove all but one frame around the edges of the trees. Again, this looks like a really good idea. Thanks

[@contrapposto8389]

Extremely well explained. Thank you

[@pretenderfishing]

Great video. I use image averaging a ton for night images, but hadn't thought that the same premise works for shadows in daytime images. I'm wondering if some hybrid of image averaging with exposure blending might be the best of both worlds.

[@frankf9233]

Very interesting! I will have to give this a shot. Thank you Spencer!

[@PhotographyLifeChannel]

Sure thing! Glad you liked it.

[@MOAB-UT]

Great tips for capturing moving. I use exposure bracketing which is what I think you are talking about. Do you know if AFFINITY Photo would let me do image (aka exposure) averaging? Thanks!

[@csc-photo]

Excellent presentation skills & helpful info, thanks Spencer! Question - how does a single 14-bit RAW image compare to image averaging, in terms of noise & range? Not sure if you're shooting RAW in this vid (forgive me if I missed this detail).

[@PhotographyLifeChannel]

Glad you liked it! I didn’t mention it in the video, but yes, I always shoot RAW. It’s usually possible to get everything right in a single shot at that point, but in the highest-contrast situations, either HDR or image averaging is still very helpful for improving your ability to recover shadows.

[@thegroove2000]

I will try this thanks.

[@joseotoya2102]

Fantastic tutorial Spencer! Thank you.

[@PhotographyLifeChannel]

Sure thing! Glad you liked it.

[@leond123]

Hi Spencer, thanks for this video. I'll often bracket for a range of exposures and then select the one that's easiest to work with. I have never heard of image averaging. II'll definitely give it a try. When you state better results, I assume you're referring to the noise (or lack of) in the shadows? One of my questions was exactly what Ricardo asked. Why not just make copies of the original image. Thanks for the explanation. Do you always use a tripod when image averaging? Or are you able to get satisfactory results shooting hand held?

[@PhotographyLifeChannel]

Yes, the better results come from the fact that image averaging reduces noise. The more photos you average, the more improvement you get. You can’t simply make copies of the original image, because averaging those copies just gets you back to the original image! The averaging process only reduces noise because the pattern of noise is different in every photo, while the “pattern of subject” stays the same. Averaging cancels out some of the noise, while leaving the subject intact. As for your other question, I use a tripod 99% of the time anyway for my landscape photos. Image averaging could conceivably work handheld, but if your photos have meaningfully different framing between them, it’ll lead to blur in your details. I wouldn’t recommend it really. (Same with handheld HDR.)

[@xzsdfvdfbv]

@@PhotographyLifeChannel Intriguing... but I guess I don't understand the noise thing. I don't see how a bunch of equally exposed photos can average to having more shadow detail than any one of the original images, at least not significantly. If the information wasn't in the image to begin with.....? Edit: After reading your responses to other people, I understand the principle, and because of the whole ghost thing with pictures that move (like leaves in the wind), this might make me happier than HDR, but for real estate shots etc. I'll stick with HDR.

[@PhotographyLifeChannel]

Don’t take my word for it - try it! Take a series of underexposed photos at any ISO (same framing, using a tripod). Then average them together, recover the shadows, and compare the shadow noise in the averaged image versus any of the originals.

[@rickb6029]

You can also make a stack your photos into layers, align the layers if needed, then make the bottom layer 100% opacity, the next layer up 50% opacity, the next 33%, then 25%, etc... this will give you an average of all photos. So: 1/1, 1/2, 1/3, 1/4,... 1/n in opacity expressed as a ratio instead of percent, where n is the total number of layers.

[@PhotographyLifeChannel]

Yes! That’s a quicker way to do it if you don’t have a huge number of photos to stack.

[@kristiankaraneshev7528]

Thanks for the amazing content! Thats a great Idea, I have been using this for some of my shots. I use more hdr as I do hdr timelapses and then grab some of the frames to create a photo as well.

[@PhotographyLifeChannel]

HDR timelapses sound awesome! I just checked out your channel, very cool. I don't do many timelapses, so I could be wrong, but I don't think image averages would be a great substitute for HDR in that case. The big issue is that image averages need to be done individually in Photoshop, which takes a minute or so per frame. Fine for single images, but not ideal for 10 seconds of video!

[@calvy26]

Excellent, I'm looking forward to testing it!

[@PhotographyLifeChannel]

Awesome, I don’t think you’ll be disappointed!

[@locp]

Thanks you Spencer, what an interesting and informative video, just like all your other videos 👍

[@PhotographyLifeChannel]

Much appreciated!

[@kingbov6250]

Fab tip. Great video. Thanks very much.

[@PhotographyLifeChannel]

You're quite welcome! Glad you liked it.

[@thegroove2000]

What about bracketing?

[@PhotographyLifeChannel]

Bracketing would be a step in the process of making a traditional HDR. It’s not necessary for the image averaging method.

[@thegroove2000]

@@PhotographyLifeChannel Thanks.

[@martinchambers6930]

Thanks for this. Very interesting! What effect do you think noise reduction software like DxO Prime would have on the each image before you took the image average, given the process you describe? Would this help or make things worse?

[@PhotographyLifeChannel]

I would rather do the noise reduction afterwards. If you did it beforehand, I’m concerned that you’d exaggerate some of the noise reduction artifacts.

[@HmS1856]

Thanks for this, Spencer. I have seen a similar technique used to remove tourists from a scene. Just to clarify, do the images you take to create an average, all have the same exposure? Also does the resultant image have a higher resolution than a single image? Thank you.

[@PhotographyLifeChannel]

Correct, the images should all have the same exposure - specifically, they should have an exposure that doesn't blow out any highlight detail at all, even if the photos seem too dark. The process of image averaging doesn't brighten them, but it minimizes noise so much that you can manually brighten them with things like the shadows slider to get excellent results. As for your second question, there is no improvement in resolution with this method, just in noise performance. However, there is a related method which can boost resolution if you shoot a burst of shots handheld (though I don't use it personally): petapixel.com/2015/02/21/a-practical-guide-to-creating-superresolution-photos-with-photoshop/

[@Cuervo79]

Would be interesting if you did a video of editing an averaged photo... To see any other benefits of using the technique, I normally use a 7 exposure HDR so that would mean like more than 16 photos for average no? but I would like to see some results of what you think it helps (not only the ghosting artifacts and noise)....

[@PhotographyLifeChannel]

The biggest benefit compared to HDR is certainly the lack of ghosting, but it does also make the editing process a bit more "subdued" for lack of a better word. It's basically like editing a single RAW photo, except one with much more flexibility than normal. I don't know that there are too many other benefits of image averaging over HDR, although image averaging definitely has benefits over single images, too. The lower noise is the biggest one, but there's also going to be less color shifting when recovering shadows.

[@GusMcCrae01]

Excellent information.

[@tonymckeage1028]

Great Video, New Sub here! thanks for sharing

[@Henry_Churches]

If you’re photographing a moving subject such as a tree, how would not still get that ghosting effect by taking multiple images? You’re ditching HDR because it requires you to take and blend multiple images, but then with your technique, you still take multiple images. I get the noise concepts, but still doesn’t seem like a solution to ghosting. Do I understand correctly that when it comes to motion, the only difference is ‘artifacts’ with HDR vs a slow shutter speed type ‘blur’ with your technique? Interesting technique either way!

[@PhotographyLifeChannel]

That’s basically right, it’s a difference between getting a long exposure-style blur with my method, compared to various digital artifacts with traditional HDR. There’s no method that totally freezes all the motion, so it’s down to which side effects you prefer - personally, I strongly favor a bit of long exposure blur! But that’s up to each photographer.

[@Henry_Churches]

@@PhotographyLifeChannel Got it! Thank you for clarifying! Will have to give the technique a try!

[@olegaertner]

Sounds great. But why is noise improved when taking four pictures at the same exposure? Shouldn't the noise be just as visible as with a single photo? I don't get the process behind it. Can you explain?

[@PhotographyLifeChannel]

The reason is that noise is random, and it completely changes position from photo to photo. So, when you average together photos, the noise cancels out while keeping the subject intact. You can think of it like averaging out coin tosses - you may start with 100% heads, but with enough flips, it will smooth out to 50/50. Likewise, one pixel may be too bright because it has a lot of noise, but the next photo is unlikely to have the same issue, or may even be too dark. The more photos you take, the more "smoothing" occurs, and the less noise you see. Test it out yourself, and you'll see how well it works!

[@shamanbeartwo3819]

@@PhotographyLifeChannel - Thank you for explaining this in a way that even I can grasp!

[@Powerstroke98]

I guess we're all different, in that seeing the two 'tree' images, I like the HDR image better!

[@howardholtzman1331]

Thank you for the great video. How do you setup your camera so you don't get any over exposed images or do you just use manual to shoot each exposure?

[@PhotographyLifeChannel]

Good question. I always use either aperture-priority mode or manual mode, which makes it easy. In aperture priority, I dial in negative exposure compensation to the point that nothing in the photo is blown out. In manual, I target the same point just by shifting the settings manually. I’ve shot with my camera long enough that I usually know what exposure compensation I need just by looking at the preview in live view. But if that’s not precise enough, you can always check your histogram or blinkies after you’ve taken the photo just to make sure that none of the highlights are blown out. (Some cameras have a live histogram while you’re taking pictures, but I don’t like covering up so much of my screen, so I keep it off.) Hope this helps.

[@howardholtzman1331]

@@PhotographyLifeChannel Thank you for the response. Please clarify, are you in Manual Mode taking each individually?

[@PhotographyLifeChannel]

@@howardholtzman1331 Again, sometimes I’m in manual, and sometimes aperture priority. Either way, I just set my camera to take a burst of photos, then fire off all 4-8 pictures for the image average at once. It’s not an especially difficult technique. You just take a series of photos in a row, all with the same brightness (and no overexposed highlights). If you don’t know how to do that in either manual or aperture priority mode, I’d recommend learning more about camera settings in general before diving into topics like image averaging.

[@SidneyPratt]

Thanks.

[@DiyEcoProjects]

Hi there, many thanks for your valuable video. Can i ask a few things please? You say i cant get back detail from highlights. So just want to check, i can take under exposed photos and draw data back out of them? This would mean that i can use ND filters and even though they come out as underexposed i can bring that back in photoshop. Sorry learning new things. All the best, kieron p.s. didnt know you could photostack in photoshop, so thats new too

[@PhotographyLifeChannel]

I don’t mean to suggest that you can have wild underexposure and bring back the shadows no matter what. Normally, that’s a recipe for huge levels of noise/grain in the shadows. Image averaging, though, can improve shadow noise by several stops, assuming you took enough photos in the field to stack. So, if you average something like 4-8 different photos together, the result has cleaner shadows, and you can recover those darker areas pretty well, within moderation. In your ND filter example, it would only be possible to bring that back in Photoshop if you have a series of those photos in a row that you can average together. If it’s just a single photo that you’re working with, there’s no way to bring back that detail, and image averaging won’t do anything.

[@DiyEcoProjects]

@@PhotographyLifeChannel Many thanks for your reply. Ahh see... so keep everything within 1-2stops of a normal exposure and do several photos. Ok ill give that a go, thanks very much. I find photoshop fascinating. Ive been playing around with different layers to get 20% mystical blurred effects. All the best on your channel, kieron

[@Ricardo-SW]

So what exactly are you doing with Image Averaging? You turned off what I assume are the HDR over/under settings, set -0.7 stop compensation and did what, take 4 shots at the same aperture/speed/iso, and then blended in PS? If the 4 are the same, why won't one do and just make duplicates if you need multiples for PS? I know you can use multiple images to reduce noise thru process you seem to be using, but how does that help the darks? Is it that you can now crank up the shadows without getting noise? I'm lost here :) Thx

[@PhotographyLifeChannel]

I know you say you’re lost, but you pretty much nailed it. I took a series of 4 photos with identical settings, slightly underexposed to avoid any blown highlights. And then because image averaging significantly decreases noise, it’s far easier than usual to bring back shadow detail. Couldn’t do it with just making duplicates of one image - naturally, averaging four copies of the same photo gets you back to the same photo! It doesn’t reduce noise. Image averaging works because the pattern of noise is random and different in every photo, so it cancels out upon averaging.

[@Haixu_Wang]

Hi, do you edit the x nr. of photos in LR first and then average them or do you average them and then edit the photo?

[@sfinnera1]

Hi there, I understand from your video to do this technique you need to ensure there is no clipped highlights. At 7:27 there is an image of a sun star, how did you process this image? Thanks

[@PhotographyLifeChannel]

Good question! It’s not exactly that you must avoid any clipped highlights at all - just that image averaging won’t help you recover them any better than before. If it’s something like the sun, which you don’t *need* to recover, then it’s perfectly fine to clip those highlights. In this particular case, I just did a bit of underexposure to make sure the blue sky didn’t blow out, but I didn’t concern myself at all with the sun itself. Then in post-processing, I recovered the shadows to get a brighter overall look, and left the sun as-is (even thought it’s clipped - I don’t mind :)

[@amirnaghib]

Thank you Spencer for the very useful videos you provide us. May I ask an irrelevant question to the topic of this video just popped up in my mind? when we use a wide or telephoto lens shooting a landscape e.g. a mountain and its surroundings is it possible to use an aperture like 5.6 which does not give you a great DOF and have a back to front sharp photo?

[@PhotographyLifeChannel]

Good question. It strongly depends on how far away the surroundings in question are. If you're referring to other distant mountains, and nothing in your photo is nearby, then it's absolutely possible. But if the surroundings are things like trees nearby you, framing the mountain, then f/5.6 isn't going to give you enough depth of field. Either the mountain or the trees will be noticeably out of focus upon close inspection. And the longer your lens, the more true this is; telephotos have significantly less depth of field than wide angles. If you haven't already seen it, my video on the best aperture for landscape photography might clarify this further: kzbin.info/www/bejne/kImvnmpjetCMias

[@LeeDHwang]

Does it matter whether you brighten before you image average in PS, or is it better to brighten the image after averaging? I tried both and it doesn't seem to make a big difference. I had thought brightening the RAW file and making any other basic exposure adjustments before exporting to PS would have been theoretically cleaner before the conversion to TIFF, but it doesn't seem to matter. Any advice would be appreciated.

[@PhotographyLifeChannel]

It won’t really make any difference!

[@omnirhythm]

That intro is super deep and meta, be it on purpose or not :D

[@PhotographyLifeChannel]

Ah yes, the light at the end of the tunnel. True in photography, and in life. [Cue eyerolls]

[@Laladeduh]

Another winner! Thank you.

[@PhotographyLifeChannel]

Much appreciated!

[@JohnnyBoyUploads]

As the photo sequence contains underexposed images, the Image Average output photo you have shown has the shadow areas brightened. Does the Image Averaging brighten it up or have you done it?

[@PhotographyLifeChannel]

Good question, you're right - the average of those underexposed shots will also look underexposed. The benefit is that the averaging method reduces shadow noise (AKA increases dynamic range) so much that recovering the underexposure is much easier than normal. I didn't show the brightening step each time in this video, but it did happen.

[@JohnnyBoyUploads]

@@PhotographyLifeChannel oh that’s great! But this is something to consider for landscape photography! Thanks for ur input! Came here from diyphotography.. grt channel!

[@PhotographyLifeChannel]

@@JohnnyBoyUploads Oh nice, DIY Photography is awesome! Welcome to the channel.

[@RewDowns]

Would you edit the raw first then edit on the tif in photoshop?

[@PhotographyLifeChannel]

It doesn’t matter for image quality, so I recommend averaging the fully unprocessed images into a TIFF as step one, so that you don’t bake any of your edits into the TIFF file.

[@shubhayu69]

Plz mention how many photos have to be captured in microthird camera for 3 stop shadow recovery Thanks in advance

[@PhotographyLifeChannel]

It’s the same with any camera - eight photos will give you three additional stops of shadow recovery.

[@shubhayu69]

@@PhotographyLifeChannel thanks If i shoot with with om1 with live ND 64 Can I get 5 stop shadow recovery??

[@caneestudio]

Phase 1 frame averaging - precision

[@davewallace5008]

But you are still taking several photos and blending them into one, I don't see the difference apart from the ghosting or am I missing something?

[@PhotographyLifeChannel]

Basically the “pattern of noise” (which is responsible for bad shadows when you recover very dark photos) changes from picture to picture, because noise is random. Whereas the “pattern of subject” stays the same, for obvious reasons. So if you take several photos and average them, the pattern of noise averages out, and the shadows become less noisy. This lets you do strong shadow recovery, AKA boosting your dynamic range as much as an HDR. (This averaging process is actually how some phones like Google Pixel create their so-called “HDR” photos, apparently.)

[@biggsy..215]

The problem i have i use Luminar and not Photoshop.

[@PhotographyLifeChannel]

Unfortunately I don't think Luminar has image averaging. But some free programs like GIMP do. Maybe you can find one that works for you.

[@osomaligno]

Great content!

[@PhotographyLifeChannel]

Thank you, Paul! Glad you liked it 😄

[@franzposch3037]

great idee

[@carljarvinen1189]

Does this work well for moving water situations like waves or a waterfalls?

[@PhotographyLifeChannel]

Yes, extremely well! That’s pretty much why I started using image averaging. I was frustrated with how HDR tends to fail in those situations, even with anti-ghosting correction turned on. The one caveat is that the faster parts of the water will be somewhat blurred with this method - not much, but about like using a 1/2 to 1 second shutter speed.

[@mugwump6400]

I don't get it: What is the point of taking multiple images with the exact same exposure settings? Can't you just apply the image-averaging technique by copying the same image multiple times and then averaging it with itself?? I somehow think, I am missing the point here...

[@PhotographyLifeChannel]

I like where your head's at, but the method you describe doesn't work. The reason why image averaging reduces shadow noise so much (AKA increases dynamic range so much) is that the pattern of noise is random - so, it changes from photo to photo. Whereas the "pattern of subject" stays in the same place. So, averaging multiple photos will cancel out a lot of the noise, while keeping the subject intact. The more photos you average, the more noise gets canceled out. It won't work if you just copy the same photo a bunch of times, because then the average would just get you right back to the same photo :)

[@poweruser64]

You know what's cool about this? This is exactly what Google's HDR+ algorithm works, what helps gives Pixel photos their look!

[@a.keithclarke7975]

But if you look at your comparisons, the HDR had far more detail.

[@zeljkoplavsic784]

Looks like photo bracketing to me

[@PhotographyLifeChannel]

There is no bracketing involved, each image is the same exposure.