Noise gathering?

Discussion forum for Tawbaware's TuFuse and TuFuse Pro software
dsjtecserv
Posts: 551
Joined: Sat May 05, 2007 3:25 pm
Location: Northern Virginia
Contact:

Post by dsjtecserv » Thu Apr 24, 2008 10:37 am

maxlyons wrote:
I'm not sure there aren't situations where this might be an intentional decision. In the case at hand, the light was fading and even the small breeze resulted in subject motion. I'm not sure I could have gotten useful images at a lower ISO.
I see the allure of doing this, but if the breeze was sufficient to cause subject motion, then won't you end up with misalignments during the exposure blending stage? If the breeze was such that an exposure at ISO 100 was not feasible, wouldn't it also preclude taking 2 (or more) exposures at higher ISOs and expecting them to line up correctly? Or, was it the case the breeze would stop for long enough to allow a X second exposure, but not a 2*X exposure, and during the moments when the breeze stopped, the subject returned to *exactly* the same position every time so that exposure blending was feasible?
Max
Yes, alignment was an issue, but Photoshop auto-align layers actually did quite well. Although there was subject movement relative to the background, this was the top of a single flower stalk so there was little or no movement of the parts of the subject relative to each other. I did wait for relatively calm periods, knowing that the position of the flower might not be the same in each shot but that the flower would be fairly still and the individual parts would not be blowing around much. Since the background was out of focus in all shots, I didn't care about the position of the flower in relation to that. Photoshop successfully aligned the flowers even though the background wasn't well aligned. You can see some loss in fidelity from the single image to the TuFused final, but it wasn't significant at a normal viewing size. (But the noise was!). If I had used a signficantly slower shutter speed the flower itself would have been at least somewehat blurred in all or most of the images.

Having said that, I agree that you shouldn't plan on rewriting TuFuse to deal with this particular set of circumstances. I think a general improvement in noise filtering would be great; the few cases where ISO 1600 is an intentional choice will still benefit.

Dave

dsp
Posts: 586
Joined: Tue Feb 20, 2007 11:09 am
Location: Hoboken, NJ

Post by dsp » Thu Apr 24, 2008 11:37 am

If we are talking noise reduction, I think more generally useful than adding any fuse/stack ordering (which can be done already, albeit with extra clicks) would be to add normal stacking, like average, min, max, median,.... Admittedly, I can use some free solutions (IRIS, imageJ), or commercial solutions (Matlab, Photoshop CS3 extended) to do these things but I feel that it is a natural feature for a program like TuFuse (or at least the pro version). Max, you've already gotten requests for an stacker that can handle 16-bit images, and you have an interface that would be well suited to this kind of thing, so how about it?

cheers, Darcy

maxlyons
Posts: 3649
Joined: Fri Jun 20, 2003 8:55 pm
Location: USA
Contact:

Post by maxlyons » Thu Apr 24, 2008 7:56 pm

dsp wrote:would be to add normal stacking, like average, min, max, median...so how about it?
I'll consider this. Of course, the sort of "stacking" that can be done in most editors doesn't perform the pyramid operations that TuFuse does. However, it should be possible to perform this type of simple stacking in TuFuse by setting the number of pyramid levels to 1.

Max

dsp
Posts: 586
Joined: Tue Feb 20, 2007 11:09 am
Location: Hoboken, NJ

Post by dsp » Fri Apr 25, 2008 12:10 am

maxlyons wrote:. Of course, the sort of "stacking" that can be done in most editors doesn't perform the pyramid operations that TuFuse does.
Max
That's certainly true, but simple stacking is sometimes nice! I know feature creep is always a problem, so thanks for considering it.
cheers, Darcy

rickb
Posts: 140
Joined: Fri Jul 11, 2003 9:20 pm
Location: SF Bay area
Contact:

Post by rickb » Sun Apr 27, 2008 6:50 pm

Noise for hdr combining in Tufuse PRO using raw images...

It seems that Tufuse Pro does create more noise compared with CS3 HDR choosing the default no noise reduction in Tufuse Pro and using raw images. Doing the same thing but using jpg images results in less noise.

This is for ISO100 1dsmk2, raw images, more than 8 stops of exposure range in 7 images, so the noise is low to begin with for mid range pixels for each input image. I'm using the default settings of Tufuse Pro except using exposure blending only.
It looks like it is using 1 iteration and 10 levels.

I need to try more options...
Anybody else see this with raw images?

thanks,
- Rick

dsp
Posts: 586
Joined: Tue Feb 20, 2007 11:09 am
Location: Hoboken, NJ

Post by dsp » Sun Apr 27, 2008 7:39 pm

When you say raw images, how are they being converted in each? Are you using dcraw for TuFuse and ACR for CS3, or something else, converting to tiff first, and using the same images as input into both?

I don't know the details of the CS3 algorithm, but it must do some blending/mapping of the original images to create the 32-bit float HDR image, and then do some pretty aggressive mapping to convert the HDR image back to a 16 or 8-bit one. Which method are you using under CS3? TuFuse operates differently of course, and goes directly to the tone mapped final image.

Can you post a crop of the two images in question?

cheers, Darcy

rickb
Posts: 140
Joined: Fri Jul 11, 2003 9:20 pm
Location: SF Bay area
Contact:

Post by rickb » Mon Apr 28, 2008 9:43 am

I was using dcraw.
To me, that is an unknown entity.

dsp
Posts: 586
Joined: Tue Feb 20, 2007 11:09 am
Location: Hoboken, NJ

Post by dsp » Mon Apr 28, 2008 10:33 am

Hey Rick,

If you feel like it, use ACR to convert the raws to tiff (or use dcraw) and compare the two methods then. I'd be surprised if CS3 gave noticeably smoother results. You could try different dcraw settings as well...

cheers, Darcy

rickb
Posts: 140
Joined: Fri Jul 11, 2003 9:20 pm
Location: SF Bay area
Contact:

Post by rickb » Mon Apr 28, 2008 7:19 pm

From a first look, it seems to be the dcraw options.
If you are doing an hdr type of combine, many of the dcraw options create very noisy results, perhaps they are leveling the histogram on each input image.
The best setting so far is to choose for highlight recovery, leave unclipped.

any other comments or results people have found?

WWG
Posts: 92
Joined: Sun Nov 26, 2006 11:53 am
Location: Maine
Contact:

Noise Reduction

Post by WWG » Tue Jun 29, 2010 10:54 pm

I'd like to resurrect this old thread about noise reduction in TuFuse because I just ran into a similar issue. I recently toured Lehman Caves in Great Basin National Park in Nevada. These caves have some artificial lighting to illuminate the cave structures so I decided to try to use the cave's lighting and avoid on-board flash. I set the ISO to 3200 and took bursts of 6 to 8 shots hand-held. I initially processed the best 5 (least motion blur) of each in Photomatix Exposure Mode (I only have the lite version and it only takes 5 images max.) and was suprised at how well they came out. The noise reduction was significant. After a pass through Neat Image, the images were clean and retained detail. They looked like any other low ISO shots.

I much prefer PTA-embedded Tufuse exposure blending outputs to Photomatix exposure blend mode for HDR so I tried running the same groups of noisy images through PTA_Tufuse and found the same problem discussed in this thread. The noise is not reduced, it may even be enhanced. I gather from this thread that the problem is the focus blending function looks for sharpness and noise is sharp.

If I buy TuFuse Pro can I force it to only do exposure blending (and noise reduction)? I have the trial version but I haven't yet learned how to write script for it. Can the focus blending be bypassed?
Werner

TerryGB
Posts: 33
Joined: Wed Mar 19, 2008 10:07 am

Re: Noise Reduction

Post by TerryGB » Wed Jun 30, 2010 4:21 am

WWG wrote:If I buy TuFuse Pro can I force it to only do exposure blending (and noise reduction)? I have the trial version but I haven't yet learned how to write script for it. Can the focus blending be bypassed?
Have you Max's tutorial on how to use TuFuse Pro?
http://www.tawbaware.com/tufusepro_help ... o_help.htm
Just select exposure blend at the top left of the configuration screen. Using just exposure blending you will find that noise is quite markedly cancelled out even without using the built-in noise reduction, personally I deselect the noise reduction.
http://www.tawbaware.com/tufusepro_help ... onfig1.jpg

maxlyons
Posts: 3649
Joined: Fri Jun 20, 2003 8:55 pm
Location: USA
Contact:

Re: Noise Reduction

Post by maxlyons » Wed Jun 30, 2010 7:39 am

WWG wrote:Can the focus blending be bypassed?
Terry is right. Focus is blending is one of many different types of blending that TuFuse and TuFuse Pro can perform, all of which is pretty extensively documented in their web pages and help files (here, here and here).

Max

WWG
Posts: 92
Joined: Sun Nov 26, 2006 11:53 am
Location: Maine
Contact:

Post by WWG » Wed Jun 30, 2010 9:38 am

Thanks Terry and Max,

I didn't understand the consequences of the second sentence under the Auto Blend discussion. Since all of my images were the same exposure Auto Blend chose to do focus blending. I reprocessed some of the ISO 3200 cave images in TuFuse Pro "exposure blend" mode and they look great. I also had some handheld multiple high ISO shots of Las Vegas at night and they came out clean as well. I wasn't quite as steady as I had hoped to be so they aren't quite as sharp as I would have liked but the averaging process certainly makes high ISO images usable.

Thanks again,
Werner

maxlyons
Posts: 3649
Joined: Fri Jun 20, 2003 8:55 pm
Location: USA
Contact:

Post by maxlyons » Thu Jul 01, 2010 9:33 am

WWG wrote:I didn't understand the consequences of the second sentence under the Auto Blend discussion.
Can you point out the sentence in question? I'm not sure which sentence you mean.

Thanks,

Max

WWG
Posts: 92
Joined: Sun Nov 26, 2006 11:53 am
Location: Maine
Contact:

Post by WWG » Thu Jul 01, 2010 4:23 pm

Auto blend. TuFuse Pro automatically determines if the images should be focus blended, exposure blended, or a combination of the two. Images that share the same (or very similar) exposure levels are assumed to be intended for focus fusion. Images with different exposure levels are assumed to be intended for exposure fusion.
It is clear in hindsight. I thought I was doing neither exposure blending nor focus blending, just averaging down noise. Since my images were all the same exposure, TuFuse assumed focus blending. I didn't realize until searching for this thread that focus blending didn't average down noise also. I first tried the PTA-embedded TuFuse and got a noisy image and then mechanically tried it in the trial TuFuse Pro without changing anything from its defaults. Once I took TuFuse Pro out of Auto mode and into exposure blend mode the noise averaging works great. I averaged six handheld, 1/8 sec, ISO 3200 images of the Las Vegas Eiffel Tower at night from my Nikon D90. The individual images are pretty ugly but TuFuse's output is nearly noise free. Neat Image's noise profile of the combined image shows very low values.
Werner

Post Reply

Who is online

Users browsing this forum: No registered users and 1 guest