Luminous Landscape Forum
Equipment & Techniques => Medium Format / Film / Digital Backs – and Large Sensor Photography => Topic started by: jonathan.lipkin on May 24, 2011, 11:32:06 pm
-
Hey everyone. You may be familiar with my ongoing search for a digital camera system with some movements. I have it narrowed down to a Rollei XAct with a Phase One back and a Hasselblad hd3 with an hts adapter. Took the Hassy out for some testing this weekend. I haven't processed all the images yet, but on some, but not all, of the images with extreme shift, there is a noticeable color cast. Starts as green at the top, the middle is normal, then the bottom is magenta.
OK, I know what you're saying - don't use the shift. But, the manual specifically describes using shift of up to 18mm in each direction. The Hasselblad tech I spoke to thought this would be corrected in Phocus, but it doesn't seem to be doing so. I have an email in and will let you know what they say, but would be interested in hearing if anyone else has the same experience. I suppose I could shoot a white diffuser over the lens to create a correction map, but seems a pain to do.
-
... I suppose I could shoot a white diffuser over the lens to create a correction map, but seems a pain to do.
Hi Jonathan,
You should indeed use a color cast correction. That's what it's for, to compensate for color cast (and vignetting and light fall-off).
Cheers,
Bart
-
Hi Jonathan,
You should indeed use a color cast correction. That's what it's for, to compensate for color cast (and vignetting and light fall-off).
Cheers,
Bart
+1
Capture One and RawTherapee (the current dev builds) can both perform this type of flat field correction.
-
Hi Jonathon,
It is pretty easy in Phocus to use Scene Calibration to correct for this.
Phocus will automatically correct for Vignetting, CA and Distortion, but there are too many variables to accurately remove casts without using the Scene Calibration tool.
Hope that helps and that you enjoyed using the HTS.
Also you can easily place your H3D or H4D on the Rollei xact, fully compatible with manual, or electronic shutters from Schneider or Rollei.
David
-
Thanks, everyone. Scene correction tells me that the image is too complicated.
-
and a Hasselblad hd3
Which one did you use? 31? 39? 50?
Thank you.
Barkeeper
-
Metadata tells me h3dII-39MS. Don't think I enabled any of the MS capabilities, but couldn't find a manual for it.
-
Thanks, everyone. Scene correction tells me that the image is too complicated.
Did you capture the scene through an Opaque filter as per any other MF system?
MS capabilities are enabled when shooting tethered to a computer. Look in the Camera Control tool to switch between Single and Multi.
I guess you are buying second hand?
David
-
"Opaque" filter? Meaning no light passed through the filter?
Jerry
-
"Opaque" filter? Meaning no light passed through the filter?
Jerry
LOL - you mean a lens cap?
David - No, I did not use a filter. Had heard about them, but didn't realize it would be an issue. The manual and brochure had not mentioned anything about the tint.
-
You need to capture a reference by taking a shot through a translucent filter (i.e. a bit of plexiglass), Phocus (and Capture one/Leaf capture do the same thing) than uses that reference shot to remove the colour casts. You can't just use the image as the reference as Phocus has no way of knowing what is cast and what is part of the scene. Hope that makes sense. As an aside I once did a copy shot for an acquaintance (i.e. for free) and didn't feel like creating a proper copy set-up so I just side lit it and used scene calibration to even up the light fall off. Worked perfectly and made me feel marginally better about shooting for free...
-
You need to capture a reference by taking a shot through a translucent filter (i.e. a bit of plexiglass),
Yup - just got off the phone with Hasseblbad tech support and that's what they suggest.
-
LOL - you mean a lens cap?
David - No, I did not use a filter. Had heard about them, but didn't realize it would be an issue. The manual and brochure had not mentioned anything about the tint.
Whoops. Opaque really not the right word. Excuse me.
In Phocus, if you goto Help > Show User Manual then you can read more about Scene Calibration there. Otherwise drop me a message and Ill send you a tutorial.
David
-
In Phocus, if you goto Help > Show User Manual then you can read more about Scene Calibration there.
Interesting. David, were I to roughly duplicate the lighting conditions (outdoor, sunny) and lens conditions (35mm lens, HTS adapter set to x tilt and x shift), do you think I could duplicate the tint and retroactively create a calibration image?
-
Interesting. David, were I to roughly duplicate the lighting conditions (outdoor, sunny) and lens conditions (35mm lens, HTS adapter set to x tilt and x shift), do you think I could duplicate the tint and retroactively create a calibration image?
Well, it has been spoken of at times.
I am not 100% convinced it would be foolproof but worth a try.
-
Interesting. David, were I to roughly duplicate the lighting conditions (outdoor, sunny) and lens conditions (35mm lens, HTS adapter set to x tilt and x shift), do you think I could duplicate the tint and retroactively create a calibration image?
Unless the software does something strange behind the scenes, there should be no problem when you try to get the lens/focus/aperture and X/Y shift the same as when the actual shot took place. I don't think the lighting conditions matter, unless there is something uncommon(*) going on in the software, or your lighting conditions were exceptional (e.g. Sodium lighting).
(*) Common procedures in Astrophotography use a normalized colorbalance for the Flat files.
Cheers,
Bart
-
Bart, the purpose here is to even out a response imbalance across the frame (similar to vignetting, but rather a linear gradient rather than radial). This can be done per channel using the calibration image (this is how Rawtherapee does it), so is independent of white balance which is a global equilibration between the channels.
-
Bart, the purpose here is to even out a response imbalance across the frame (similar to vignetting, but rather a linear gradient rather than radial). This can be done per channel using the calibration image (this is how Rawtherapee does it), so is independent of white balance which is a global equilibration between the channels.
Hi Emil,
Yes, there are 2 possibilities. The channels of the cast correction image can be individually normalized to 1.0 from the maximum value of each channel itself, or from the maximum of all channels. The latter could introduce a White balance shift when dividing, so I assume it is not implemented to do it that way.
Cheers,
Bart
-
I just got myself a piece of white, semi-opaque plexi and have done some preliminary testing with Scene Calibration. So far, so good. It removes the green/magenta cast well. My print viewing board (which I'm shooting to test) isn't evenly illuminated enough to do super-precise testing, but the cast is very apparent in non-corrected images, and nearly completely removed in corrected images.
The 'Equalize intensity' function works less well on the more extreme shifts (+15) but there does not seem to be much intensity unevenness.
Further, I tried retroactively creating a scene calibration image and it worked pretty well. I shot my print viewing station which is lit by daylight flourescent bulbs at the HTS settings of the images I had shot a couple of weeks ago on the beach. I then created Scene calibrations from them and applied to the beach images. Not perfect, but a huge improvement.
-
=> David Grover
Hi David,
Rather than start a new topic, I thought I might simply ask the question here.
I'm Windows 10/Phocus 2.9.1
When attempting to create a scene calibration, I am sometimes getting the error "Failed to create correction. Correction error." Repeatedly clicking Create makes no difference. Some files (with the same exposure) work and some don't.
This error, as far as I can see, is not mentioned or dealt with in the Phocus Help pdf.'s.
Any idea what causes this and how to avoid it?
Many thanks.
D.
-
Dinarius -
When creating calibration files, i sometimes get an error message if my calibration images are too light, though it is not the same error message you are getting.
I was told by Hasselblad that the correct exposure is middle gray +1.5 - 2 stops
How are you creating your calibration files?
-
Hi Jonathan,
Thanks for the reply.
I'm coming to this fresh faced and I'm not in a position to do it properly over the weekend.
I read that, for the purposes of art repro, you can shoot a white card or use a translucent filter.
So, I shot an evenly illuminated white translucent window blind, just to see what would happen.
As you can see on my thread here> http://forum.luminous-landscape.com/index.php?topic=106619.0 I got a clear green/magenta cast on the images with the 120mm lens. This was totally eliminated when I succeeded in creating a scene calibration. Great! My only concern is the fickleness of the software. But, that could be down to not using a perfectly even surface (which the textured window blind isn't) for the source file. I don't know.
One question I have is this........
A typical workflow for me would be capturing single and multi on location to a laptop with Flexcolor installed. I then transfer to a desktop back at base and edit on Phocus.
Can I shoot the target file (either white card or translucent filter) on location and then create the scene calibration back at base when I'm editing? I presume I can.
Thanks again.
D.
-
So, I shot an evenly illuminated white translucent window blind, just to see what would happen.
But, that could be down to not using a perfectly even surface (which the textured window blind isn't) for the source file. I don't know.
If you shoot the blind, presumably you are quite close to it, at infinity focus you should loose the texture and more closely approximate the translucent sheet.
My, admittedly limited, experience of using correction files is the replication of the illumination "off site" is not easy, if you are using a repro set up for both it "should" be easier. The requirement for a file for each set up shows that a generic file is not good enough otherwise it would be easy and that will never do.
-
Why don't you just take something with you to take a calibration picture? I recently purchased an ExpoDisk 2.0 and it works very well and it's small enough it fit into your pocket or backpack.
-
Why don't you just take something with you to take a calibration picture? I recently purchased an ExpoDisk 2.0 and it works very well and it's small enough it fit into your pocket or backpack.
Hi,
A gentle warning against using an Expodisk for this purpose. I've seen examples of a less than perfectly uniform diffusion of light (which is not its intended use, so no critique). The closer the focus (so especially with Repros), and the shorter the focal length, the more uneven will the calibration shot look, and affect the correction based on it. Semi transparent acrylic or Perspex or opal glass is much better suited for flatfielding, and color cast removal.
Cheers,
Bart
-
My Expodisc lives in my gear bag and does double duty for setting white balance and, when the occasion has required so far, making lens cast correction files for my Hassy SWC Biogon. It's fine as long as it's large enough to cover the front of your lens (they come in different diameters) and you manage to keep your fingers out of the way. ;)
Per Bart's comment: a translucent plastic card is probably the best solution and easy to carry along in the field (and certainly less expensive than an Expodisc). At close focus and/or with short lenses you might pick up the texture of the diffusers of the Expodisc if you have the diffusers facing toward vs. away from the lens. I'll need to take a closer look to see if I can detect any issues with uniformity on the LCC files I've made with the Expodisc for the Biogon. I've been meaning to pick up a more proper tool for this, now that I've started experimenting with tilts and shifts on the Flexbody with the 40mm IF CFE (which is too big for my Expodisc in any case). With Bart's latest post I'll now move this up on my to-do list...
John
-
Hi,
A gentle warning against using an Expodisk for this purpose. I've seen examples of a less than perfectly uniform diffusion of light (which is not its intended use, so no critique). The closer the focus (so especially with Repros), and the shorter the focal length, the more uneven will the calibration shot look, and affect the correction based on it. Semi transparent acrylic or Perspex or opal glass is much better suited for flatfielding, and color cast removal.
Cheers,
Bart
But then again - if you're making repros you wouldn't go for a short focal length and closeups. I actually can't think of a single photo I ever took with a short focal length and close focus... But even if - I don't think the ED would produce uneven results, I'll test it tomorrow but I just can't see how it might happen.
Do you have the first version or version 2.0?
-
But then again - if you're making repros you wouldn't go for a short focal length and closeups. I actually can't think of a single photo I ever took with a short focal length and close focus... But even if - I don't think the ED would produce uneven results, I'll test it tomorrow but I just can't see how it might happen.
Do you have the first version or version 2.0?
Hi Christoph,
The closer focusing distance doesn't help either. I do not use an Expodisc myself, but I've seen examples of the gridpattern interfering with what a diffuser for Flatfielding or ColorCast removal should do, i.e. produce a completely featureless diffuse and out of focus (because too close if flush with the lens threads) light surface.
It's not unexpected, because the device was not designed to photograph through, but to make an integrated measurement of total angle of view Exposure level and Whitebalance by removing most detail.
I can't find the examples on the web right now, but I know they threw some efforts that needed a fully diffuse light surface off, enough to cause problems with ColorCast removal and light fall-off correction.
A diffuser on the other hand should mimick a Lambertian Reflection of a perfectly uniformly lit featureless surface, a uniform lightfield. So it should be as insensitive to the angle of incoming light as feasible, and reflect or transmit perfectly diffuse in all directions. Transmission is easiest, because that avoids reflections from the rear surface, but an out of focus or featureless diffuse reflecting surface also works (although it's harder to illuminate a larger surface uniformly).
Cheers,
Bart
-
FWIW, I just checked the LCC files I made in Capture 1 using an Expodisc on the Biogon 38mm (shot on an IQ160, focused at infinity or near-infinity and stopped down to f/11 or f/16). I looked pretty carefully and don't see any rippling either in the LCC file itself or in the corrected files, but YMMV as they say... I think the bottom line is that if you have an Expodisc already, it's worth trying as a LCC plate. If not, a translucent piece of plastic is probably a safer and cheaper way to go.
-
I've been using 1/8x5x6" translucent white acrylic "sign lighting white 40%" from Tap Plastics for several years. You can get several cut for $10.
http://www.tapplastics.com/product/plastics/cut_to_size_plastic/acrylic_sheets_color/341 (http://www.tapplastics.com/product/plastics/cut_to_size_plastic/acrylic_sheets_color/341)
-
FWIW, I just checked the LCC files I made in Capture 1 using an Expodisc on the Biogon 38mm (shot on an IQ160, focused at infinity or near-infinity and stopped down to f/11 or f/16). I looked pretty carefully and don't see any rippling either in the LCC file itself or in the corrected files, but YMMV as they say... I think the bottom line is that if you have an Expodisc already, it's worth trying as a LCC plate. If not, a translucent piece of plastic is probably a safer and cheaper way to go.
That's right, better safe than sorry. There is one issue that makes it harder to detect on an outputfile though. The LCC data is calculated in linear gamma, probably before demosaicing. When the gamma precompensation is appied to the already demosaiced data, it compresses the highlights in anticipation of the display gamma decompressing it again. Subtle differences can get lost there.
I know from my own experiments, and from experiments done by Jim Kasson and Jack Hogan for the extraction of Photon Transfer Function curves from Raw files, that a careful field flattening can reveal very subtle irregularities in the photosite responses. Maybe it was one of them who discovered the Expodisc effects, don't recall for sure.
Cheers,
Bart
-
Can I shoot the target file (either white card or translucent filter) on location and then create the scene calibration back at base when I'm editing? I presume I can.
That's my workflow - shoot a calibration image with thin plexi over the lens in the field, create a calibration file when I get back to the studio. I shoot with an HTS adapter, which creates the cast. For the times when I forget or don't have time to shoot a calibration image, I just shoot a calibration image with the plexi with the hts set to the same tilt and shift of the original image.
With the HTS the color cast is caused by the light hitting the sensor at an angle, from what I understand, so if I duplicate the angle in the studio, I get the same cast. I'm not sure why you are getting the cast with your lens. But, just try a few tests.
PS, each calibration file you create is an xml file in ~/Library/Application support/Phocus/Settings/User defined/Scene calibration. If you want to use the files on another computer, just copy the contents of that folder
-
BartvanderWolf - my apologies!
I just tested and although under normal usage there is absolutely no sign of the "rippling", you can actually see them under certain circumstances. So with the 50mm at lose focus it's slightly visible - as with a 120mm macro at 1:1.
Though as I said - under normal circumstances it's not at all visible.
-
The Phocus scene correction works well. It's been years since I've shot with a 39 MS. But I certainly used the tilt-shift gizmo outdoors and was able to use the scene correction function to compensate for color, falloff, and optical distortion. The 50mm is a dud, but the 50mm II is excellent.
-
The Phocus scene correction works well. It's been years since I've shot with a 39 MS. But I certainly used the tilt-shift gizmo outdoors and was able to use the scene correction function to compensate for color, falloff, and optical distortion. The 50mm is a dud, but the 50mm II is excellent.
I shoot with the 50 frequently and like it - what did you find lacking?
-
Thanks all for the feedback.
Well and truly sorted now, I think. 8)
The Scene Calibration tool is excellent.
D.