The GetDPI Photography Forum

Great to see you here. Join our insightful photographic forum today and start tapping into a huge wealth of photographic knowledge. Completing our simple registration process will allow you to gain access to exclusive content, add your own topics and posts, share your work and connect with other members through your own private inbox! And don’t forget to say hi!

Fun with Frame Averaging Images

dougpeterson

Workshop Member
I thought I'd start a thread that honors the tradition of the prolific Fun with MF Images and the Technical Camera Images threads.

I think the availability of native/simple/fast/easy/gapless/in-camera/raw-file-creating Frame Averaging will completely revolutionize landscape and architectural photography. Because it's a new way of approaching exposure it's hard to know how to best use it, both technically and aesthetically. It raises questions like:
- what kinds of scenes does it work well in? When does it not?
- What per-frame shutter speeds are best in what situations?
- When is a normal long exposure a better option?

I'm hoping this thread can foster the kind of sharing and discussion that helps the community approach these questions and create an open resource for those looking to explore Frame Averaging in their work.

For an overview of frame averaging as implemented on the Phase One IQ4 see Phase One Frame Averaging. If other cameras add this kind of feature I encourage others familiar with those cameras to post relevant links/tutorials for those cameras as well.

I expect to contribute my own first images to this thread this weekend!
 
The article doesn't really suggest the advantage of Frame Averaging other than noise reduction. I am struggling to imagine how I would make use of it. I won't address the absurd claim that Phase One is the only camera company focused on the needs of professional and serious amateur photographers. The marketing people at DT need to tone it down a notch.
 
Last edited:

jng

Well-known member
To be fair, his enthusiasm for the new IQ4 update aside, I think Doug intended this thread to include people's experiences with frame averaging on any platform.

In terms of use cases, a friend of mine uses it to cancel out effects of convection waves that can obscure distant objects. I've seen his technique work wonders for late afternoon cityscapes in which there was quite a bit of heat radiating from ground level. In addition to obviating issues with noise when doing extended long exposures, frame averaging is also helpful when one neglects to pack the ND filters before heading out. :banghead:

Below is a comparison of one such instance (i.e., ND filters left at home) using sequential manual captures on my Hasselblad X1D w/250 Superachromat. The first image is a single 11 second exposure, the second was made from ten sequential 11 sec exposures stacked, aligned and averaged in Photoshop. All exposures were made with the electronic shutter.

Single image:
H12_26_2018_0077-16x9-small.jpg

10 images stacked and averaged in Photoshop:
H12_26_2018_70-79_mean_stack-small.jpg

A couple of observations:

- The tops of fast-moving fog smoothed out nicely; water is a bit smoother as well.

- Some highlights and light trails (from moving cars) are attenuated.

- There's a very slight decrease in sharpness detectable when viewing the full-sized file at 100%, but nothing to get too bothered about.

- John
 

dougpeterson

Workshop Member
Absolutely, any brand/make or model are welcome.

And any method of achieving frame averaging.

My only suggestion is that you explain (as jng did) how frame averaging improved the technical or aesthetic quality of the image.

If you’d like to have other off-topic conversations I would politely request you do so on a new or different thread.
 

Paul2660

Well-known member
Hello Jng,

Interesting. I assume that you are stacking, aligning, then creating a smart object and then using one of the Photoshop modes (mean or mode?)

Similar technique is what I still use for night work, star trails, as you get so much more out of the images.

Paul C
 

jng

Well-known member
Hello Jng,

Interesting. I assume that you are stacking, aligning, then creating a smart object and then using one of the Photoshop modes (mean or mode?)

Paul C
Apologies for the shorthand. After processing all the source raw files identically and outputting them as tiffs, I stacked the files in Photoshop using mean mode with auto-align box checked (File->Scripts->Statistics). Not sure I created a smart object in the process but I think not (would need to be post hoc as I don't see an option to create a smart object in the pop up window). I may have tried median mode as well - I sometimes compare processing using mean vs median and simply choose which version I prefer after seeing the result. Sometimes there's not much difference, but median mode might work better for removing moving objects like people or cars from a scene.

HeliconFocus should be able to do this as well, no?
I don't know, as I've only used Helicon for focus-stacking.

- John
 

JimKasson

Well-known member
HeliconFocus should be able to do this as well, no?
You don't get the averaging in Helicon Focus, and it may show artifacts in the presence of movement. At least, the way I use it, that's true. There may be another stacking mode somewhere that just does alignment and averaging, but I've not seen it.

Jim
 

Paul2660

Well-known member
In my experience Helicon focus cannot handle any subject movement. Creates artifacts that cannot be worked around. Which is why the XF focus stacking has limited if any use outdoors if there is wind. But after seeing some other test shots the frame averaging has just as much issue with wind created movement as a single long exposure does. So in real life work unless at Grand Canyon or similar type of setup it will be limited. Even a 1/2000 frame averaging shot had considerably harsh wind blur.

Paul C
 
Last edited:

jng

Well-known member
I mentioned earlier in the thread that one use case of frame averaging is to overcome the effects of convection or atmospheric distortion on image sharpness. My friend gave me permission to post an example of a cityscape he created from 30 images taken in rapid succession and stacked in Photoshop:

https://www.flickr.com/photos/kuriyan/40376738132/

Zoomed in:

https://www.flickr.com/photos/kuriyan/40443030971/

I think it would be a big plus to be able to do this in-camera with a single output.

John
 

JimKasson

Well-known member
In my experience Helical focus cannot handle any subject movement. Creates artifacts that cannot be worked around. Which is why the XF focus stacking has limited if any use outdoors if there is wind.
Paul C
If there's subject motion, some of the Helicon artifacts can be interesting:

2019-04-25 13-47-34 (C)-Edit.jpg

Jim
 

Audii-Dudii

Active member
Absolutely, any brand/make or model are welcome.

And any method of achieving frame averaging.

My only suggestion is that you explain (as jng did) how frame averaging improved the technical or aesthetic quality of the image.
Well, since this thread is open to all cameras, I'll post a few of my photos.

FYI, I photograph urban and suburban street scenes mostly at night, using long exposures at base ISO.

Here are a few recent examples:







I typically take seven photos (sometimes more and occasionally fewer, but always an odd number so the software won't have to compute the median values, merely select them) then median-blend them Photoshop.

Not only does this reduce noise in the shadow areas, increasing the SNR and allowing my camera (in this case, a Sony A7R mounted on a modified Cambo WDS) to punch above its weight, it also removes light trails from passing aircraft and cars, as well as stars and star trails from the sky; eliminates pedestrians that are walking through the scene; and softens and smooth clouds in the sky. When everything works perfectly, it also reduces the appearance of jaggies along sharp edges and reveals additional detail in the various surfaces and textures visible in the photo.

Of course, it also smears any tree leaves or branches that are blown by a breeze, as well as their shadows, but most of the time, I actually like this effect, as it adds a motion component to what are usually very still scenes, such as happened with some of the mesquite tree leaves in this photo:

 
Last edited:

dougpeterson

Workshop Member
I typically take seven photos (sometimes more and occasionally fewer, but always an odd number so the software won't have to compute the mean values, merely select them) then median-blend them Photoshop.[...]it also removes light trails from passing aircraft and cars, as well as stars and star trails from the sky; eliminates pedestrians that are walking through the scene; and softens and smooth clouds in the sky. When everything works perfectly, it also reduces the appearance of jaggies along sharp edges and reveals additional detail in the various surfaces and textures visible in the photo.

Of course, it also smears any tree leaves or branches that are blown by a breeze, as well as their shadows, but most of the time, I actually like this effect, as it adds a motion component to what are usually very still scenes, such as happened with some of the mesquite tree leaves in this photo:
Really lovely images; I particularly like the red-glow house image. And great explanation of the use of median frame averaging in your work.

Thanks for sharing!
 

dougpeterson

Workshop Member
I mentioned earlier in the thread that one use case of frame averaging is to overcome the effects of convection or atmospheric distortion on image sharpness. My friend gave me permission to post an example of a cityscape he created from 30 images taken in rapid succession and stacked in Photoshop:

https://www.flickr.com/photos/kuriyan/40376738132/

Zoomed in:

https://www.flickr.com/photos/kuriyan/40443030971/

I think it would be a big plus to be able to do this in-camera with a single output.

John
Very interesting! Thanks for sharing!
 

Audii-Dudii

Active member
Really lovely images; I particularly like the red-glow house image. And great explanation of the use of median frame averaging in your work.
Thanks for your kind words!

Some photographers will find the following 100% crops (taken from an alternative version of the last photo I posted) instructive.

A single-frame capture:



Seven identical files captured (including the single file used above), then median-blended in Photoshop:



I'll let everyone draw their own conclusions re: the pros and cons, but I believe these crops provide a representative example of the effect(s) that median-blending multiple files can achieve.

FYI, median-blending multiple files also works very well for reducing noise in the IR photos (which tend to be noisy under the best of conditions) I take during daylight hours, too! Here's an example of that:



And for grins and giggles, here's a 4:1 pano I stitched together from 35 individual files. First, I median-blended seven files into one panel, then repeated this four more times with the other 28 files to create a total of five panels. I then stitched the five panels together into a pano and although this was intended to be nothing more than a bit of fun while I was out walking my dog one afternoon, I think the final photo turned out surprisingly well:



As you can see, blending multiple files together works well for many purposes other than just reducing noise in long-exposure photos taken at night. :)
 

Shashin

Well-known member
I thought I'd start a thread that honors the tradition of the prolific Fun with MF Images and the Technical Camera Images threads.

I think the availability of native/simple/fast/easy/gapless/in-camera/raw-file-creating Frame Averaging will completely revolutionize landscape and architectural photography. Because it's a new way of approaching exposure it's hard to know how to best use it, both technically and aesthetically. It raises questions like:
- what kinds of scenes does it work well in? When does it not?
- What per-frame shutter speeds are best in what situations?
- When is a normal long exposure a better option?

I'm hoping this thread can foster the kind of sharing and discussion that helps the community approach these questions and create an open resource for those looking to explore Frame Averaging in their work.

For an overview of frame averaging as implemented on the Phase One IQ4 see Phase One Frame Averaging. If other cameras add this kind of feature I encourage others familiar with those cameras to post relevant links/tutorials for those cameras as well.

I expect to contribute my own first images to this thread this weekend!
This thread sound more like advertising Phase products than a legitimate thread on a technique. I really don't mind GetDPI sponsoring camera companies, but at least tag this thread as a Phase sponsored thread. Personally, I would rather see members join the Fun with MF or Technical Camera threads than having a thread devoted to a fairly specific camera technique. I really like seeing the diversity of work along side of each other. It is that diversity that makes GetDPI a very strong community. Also, threads are conversations. They can get off topic from time to time, but that is not reason to delete contributions, particularly when they are actually related to the topic--the problems of time displacement.

Now about frame averaging. Frame averaging has been around for a long time and did not revolutionize landscape photography. In camera averaging for RAW files is nice, but that can already be done in post. Phase is just making it easier. It is a compromise method because of the durations of the exposures. You are simply trading noise for temporal issues.

Naturally, what is the point of frame averaging on extremely high-resolution images? If it just for noise control, then at 100MP it is not very significant. Human perception is just not going to see the benefits. High-contrast areas of the image are the areas perceived by humans--eye tracking has shown that consistently. Low contrast areas of the image where noise can be seen is out of the area of vision that perceives at resolutions that could perceive noise and into regions of visions that will not be able to perceive that noise (assuming some extreme viewing distances).
 

P. Chong

Well-known member
I am sorry, I am confused...from what I understand this technique can allow one to leave the 10 stop ND behind...but in jng's example, the regular one shot is 11s exposure, and the averaged image is 10 images, each 11s. How does that work to allow one to leave the big stoppers at home?
 

Boinger

Active member
I am sorry, I am confused...from what I understand this technique can allow one to leave the 10 stop ND behind...but in jng's example, the regular one shot is 11s exposure, and the averaged image is 10 images, each 11s. How does that work to allow one to leave the big stoppers at home?
He chose to use a longer length of time for his averaging.

So with the phase one it would translate to 110s of exposures.

If you didn't have a ND filter instead of doing a 11s exposure you could do 110 x 1s exposure. Or any combination of the sort and it would render the same effect.

Essentially you pick the overall luminosity of the image and then you set the blend time, in terms of how much overall motion you would like to capture.
 

P. Chong

Well-known member
Ah, I see...thanks.


He chose to use a longer length of time for his averaging.

So with the phase one it would translate to 110s of exposures.

If you didn't have a ND filter instead of doing a 11s exposure you could do 110 x 1s exposure. Or any combination of the sort and it would render the same effect.

Essentially you pick the overall luminosity of the image and then you set the blend time, in terms of how much overall motion you would like to capture.
 

Craig Stocks

Well-known member
Here's a scene I shot last night using the frame averaging tool on my Sony a7r2 (the Smooth Reflection app). I choose to shoot with the Sony instead of the Phase One because it was a group teaching shoot and I don't have long lenses for the P1 and I anticipated needing some reach.

I photographed the scene in two pieces, I set the camera to average 64 frames for the water in the foreground and then immediately took a single frame to capture the moon. The two were then blended in Photoshop.

Sony a7r2 with Sony 70-200 f/4 @ 200mm, ISO 100, 1/25 @ f/9.0. 64 frames averaged for the water.

Fun fact - the mesa on the left is known as "Molly's Nipple."
 

Attachments

Top