|
LeicaPlace |
|
|
886Thanks
 |
|
|

March 17th, 2010, 05:02 PM
|
 |
Mu-43 All-Pro
|
|
|
|
Thanks Amin
I've just been given a link by Alan Wolf showing a US$50 Mac bundle of 10 Apps - Hydra is included plus Parallel - may just give it a go
Cheers
Last edited by BillN; March 17th, 2010 at 05:06 PM.
|

March 18th, 2010, 05:03 AM
|
 |
Aussie Legend
|
|
|
|
Today I was working on an Excavator in the Millmerran Power Station grounds. I took a quick few snaps while the security officer wasn't' looking as i had lunch. Turned out beautifully in HDR. This is the boiler unit. Tomorrow i hope to get a snapshot of the Generator rooms.
As you can see i like my HDR just very subtle. Just only need a little bit.
|

November 1st, 2011, 03:13 PM
|
|
|
Potato Hell
The potato harvester threw a conveyor belt (web) so i shot this scene while waiting for the farmer to fix it.
G3 with 50mm f1.8 Zuiko . Single raw file. HDR made from 4 versions
-2ev, -1ev. as shot, and +o.5ev
P1040115 (4) by Fungal infection, on Flickr
|

November 1st, 2011, 03:49 PM
|
 |
Wikipidoit
|
|
|
|
Noisy HDR?
One thing to watch for is the under exposures have to be boosted and so the noise levels are raised relative tothe shorter exposure elements in the HDR stack. One trick is to take multiple images at the same exposure level to be combined so as to reduce the noise level. Simple averaging works but there are better methods as well. This is called stacking (in this context).
Pretty straight forward to figure how much noise reduction is needed to offset the boost ... NxN number of images give N fold reduction in noise. A simple +1,0,-1 bracket means a 8 fold boost so need 64 images .... yikes! Most folks settle for quite a bit less as generic/favorite NR can be applied in addition to a mild NR from stacking.
I like to work with 1, 4 and 9 images as a set for reasonable levels of noise reduction for a managable number of images. If one finds unacceptable noise levels or artifacts from excessive NR in their HDR might want to give this frame stacking a try. Generally only the most demanding scene needs this kind of treatment.
PS I have a patent on these methods (US6909459) but pretty much unreadable. The jist is to lower the noise levels in proportion to the boost applied.
__________________
E-P1 now displaced by E-M5. GAS has passed. Critique is welcome but better make it harsh or I won't take you seriously.
My Minus Pages My lame website ...some photos
www.sporktography.com
|

November 1st, 2011, 03:59 PM
|
|
|
|

November 2nd, 2011, 06:09 AM
|
 |
Aussie Legend
|
|
|
|
I had forgotten about this thread and when I saw it pop up today it made me want to try out another HDR photo this afternoon. I lost Photomatix when i installed a new hdd in my laptop so i downloaded Nik's HDR efex pro and giving it a try. I like what i see.
|

November 2nd, 2011, 11:51 AM
|
 |
Mu-43 All-Pro
|
|
Join Date: Aug 2010
Location: Midlothian, VA
Posts: 1,224
Real Name: Richard Elliott Narnian's Gallery
|
|
This is a work in progress from my vacation last month. Need to work on the clouds to see if I can get them better aligned. Using Nik HDR Efx Pro.
__________________
Pax,
Richard
--------------
I hate cameras. They are so much more sure than I am about everything. ~John Steinbeck
|

November 2nd, 2011, 01:02 PM
|
|
|
Quote:
Originally Posted by RobWatson
One thing to watch for is the under exposures have to be boosted and so the noise levels are raised relative tothe shorter exposure elements in the HDR stack. One trick is to take multiple images at the same exposure level to be combined so as to reduce the noise level. Simple averaging works but there are better methods as well. This is called stacking (in this context).
Pretty straight forward to figure how much noise reduction is needed to offset the boost ... NxN number of images give N fold reduction in noise. A simple +1,0,-1 bracket means a 8 fold boost so need 64 images .... yikes! Most folks settle for quite a bit less as generic/favorite NR can be applied in addition to a mild NR from stacking.
I like to work with 1, 4 and 9 images as a set for reasonable levels of noise reduction for a managable number of images. If one finds unacceptable noise levels or artifacts from excessive NR in their HDR might want to give this frame stacking a try. Generally only the most demanding scene needs this kind of treatment.
PS I have a patent on these methods (US6909459) but pretty much unreadable. The jist is to lower the noise levels in proportion to the boost applied.
|
Rob,
Do you have examples from one set of images showing the difference between a simple 1 image raw, 3 image -1,0+1 std. HDR vs. HDR image stacked at each exposure 1st then doing the HDR? I'd be curious to see the difference you can personally achieve
Image 1: Simple 1 show RAW. Making of 3 exposures, then HDR.
Image 2: Simple -1, 0, +1 image HDR
Image 3: 3+ exposures at -1 stacked, 3+ exposures at 0 stacked, 3+ exposures stacked at +1 then HDR
Also, I didn't follow the math on the number of images required for xfold = 64 (NxN)
I get 8 fold increase to 64 images means 8x8.. but how did you get to the 8 fold boost if you use a 3 image bracket?? Is it simply 2x2x2 i.e. 2 to the power of the number of exposure values???
So would you then go about doing 64images/3exposure values so ~20 exposures at each exposure setting?
Does it make sense to weight each exposure setting equally or more on the under or overexposed vs. neutral setting (unequal distribution of images 40% at -1, 20% at 0, 40% at +1..) or even further offseting that distribution depending on if you were trying to preserve highlights or lowlights in the HDR...
As you can tell, I am very interested as this is one area I have not really played with but very much interested in .. 1 for noise reduction but also for cleaner more natural looking HDR-Tone Mapping..
__________________
I was always eyeing a Sigma DP2s but with too many short comings to be my only body. Finally treated myself to proper camera. An Olympus EP-2 with a kit lens. 20mm/1.7 soon to be had, and the addition of a legacy zoom. May still get a used DP2s for landscape photography.. just something about that Fovean look. If anyone has a PS action replicate it, please let me know.
|

November 2nd, 2011, 02:34 PM
|
 |
Wikipidoit
|
|
|
|
Quote:
Originally Posted by LovinTheEP2
Rob,
Do you have examples from one set of images showing the difference between a simple 1 image raw, 3 image -1,0+1 std. HDR vs. HDR image stacked at each exposure 1st then doing the HDR? I'd be curious to see the difference you can personally achieve
Image 1: Simple 1 show RAW. Making of 3 exposures, then HDR.
Image 2: Simple -1, 0, +1 image HDR
Image 3: 3+ exposures at -1 stacked, 3+ exposures at 0 stacked, 3+ exposures stacked at +1 then HDR
Also, I didn't follow the math on the number of images required for xfold = 64 (NxN)
I get 8 fold increase to 64 images means 8x8.. but how did you get to the 8 fold boost if you use a 3 image bracket?? Is it simply 2x2x2 i.e. 2 to the power of the number of exposure values???
So would you then go about doing 64images/3exposure values so ~20 exposures at each exposure setting?
Does it make sense to weight each exposure setting equally or more on the under or overexposed vs. neutral setting (unequal distribution of images 40% at -1, 20% at 0, 40% at +1..) or even further offseting that distribution depending on if you were trying to preserve highlights or lowlights in the HDR...
As you can tell, I am very interested as this is one area I have not really played with but very much interested in .. 1 for noise reduction but also for cleaner more natural looking HDR-Tone Mapping..
|
For frame stacking to obtain noise reduction the rule is N fold reduction for NxN frames. This comes frm the statistics and physical properties associated with counting photons.
It might help to think in terms of signal rates. For example an image with 1 second shutter if you divide all the pixel values by 1 second you get counts per second ... for a 10 second shutter divide the values by 10 and still get counts per second. Darker objects have a lower signal rate (counts per second) that bright objects. Now just add the images together!
What you see right away is the short shutter (bright objects) times result in higher noise while the darker objects (longer shutter times) bring less noise (as they are divided by a bigger number). To 'fix' this issue multiple images of the brighter objects need to be stacked (averaged) to bring the noise level down. In this example of 1 second and 10 second shutter times there is a 10x 'boost' factor so one would be tempted to shoot for a 10x noise reduction factor (requires 100 images).
In practice such exactitude and extremes are not required for getting decent resultant images (although in some cases it does-highly technical).
I hiope this is clear enough to work out a more general application for 3+ exposure brackets! Basically the longest shutter time sets the overall level of needed noise reduction.
A hybrid approach in which noise reduction (like that found in NIK or NoiseNinja, LR, etc) which trades resolution for less noise can be used in combination with frame stacking to obtain better looking images without all the overhead for full NR via frame stacking. The actual quantitative measure of NR from LR or NoiseNinja can be measured for a particular setting but sometimes going by eye is plenty.
For a +1,0,-1 bracket that is two stops so 4 fold (relative not absolute). The -1 set needs the most frames so 16 frames for 4 fold. the 0 EV set needs 2 fold so 4 frames and the +1 EV 'set' can be single frame.
Clear as mud!
__________________
E-P1 now displaced by E-M5. GAS has passed. Critique is welcome but better make it harsh or I won't take you seriously.
My Minus Pages My lame website ...some photos
www.sporktography.com
|

November 2nd, 2011, 02:42 PM
|
 |
Wikipidoit
|
|
|
|
Quote:
Originally Posted by LovinTheEP2
Rob,
Do you have examples from one set of images showing the difference between a simple 1 image raw, 3 image -1,0+1 std. HDR vs. HDR image stacked at each exposure 1st then doing the HDR? I'd be curious to see the difference you can personally achieve.
|
This sounds like an excellent example ... sadly, no I don't have such a composition to show!
Using the method I described I have used a camera with 2000:1 dynamic range and constructed a final image with 80,000:1 dynamic range with all objects of interest with at least 3:1 signal to noise ratio. The subject matter is amazingly boring set of power stabilized LEDs. Also, proprietary and I cannot share it - not much to look at in any case.
A bigger issue is just how to display the exceptional range of brightness levels in such a way on limited media like a computer screen? I don't think it can be done in a 'pleasing' and 'natural looking' way so this is where objecive and subjective clash headfirst. In such images there is way more information than the eye can see and even more than the brain can percieve. Even if we would see/percieve the 'look' of the image is so far out of our experience it will always look fake or contrived. I think this is the general knock on HDR and I'm only helping make it worse!
__________________
E-P1 now displaced by E-M5. GAS has passed. Critique is welcome but better make it harsh or I won't take you seriously.
My Minus Pages My lame website ...some photos
www.sporktography.com
|
 |
|
| Thread Tools |
Search this Thread |
|
|
|
|
|
More Discussions |
|
Click the "101 Active Discussions" tab at the top of the page.
|
|
More Member Ads |
|
Click the "Buy and Sell" tab at the top of the page.
|
|
FTC Disclosure |
This site uses affiliate programs and referral links for monetization.
|
|