Colour casts Hassleblad/P25+

[imagetone]

I have just upgraded to a Phase One P25+ using it on an H1 with 80mm, 150mm and 35mm lenses. When using these lenses straight (without tilts or shifts) the files, without any adjustments, show quite marked colour casts with all three lenses.

For example - shooting the opal lens cast calibration plate against an overcast sky and taking the grey balance in the centre - reading in a straight line across the centre of the frame (not in the corners) I get the following RGB values :  RH edge 174,171,170    Centre 176,176,176    LH edge 168, 171, 161.  The edge readings are near the edge of frame but not right at the edge.  This is the second back I have tried that has shown this problem.

Has anyone else seen colour casts of this magnitude when using normal MF lenses with the Kodak sensors used in the P25+ and similar backs from Phase, Hasseldblad or others?  How have they dealt with it?

Phase's answer is that its a feature of the chip and to use a Lens Cast Calibration on every single shot, which I find extremely surprising as all the marketing suggests this isn't needed unless using movements. And is only really any kind of answer if it can be guaranteed that only one calibration shot per lens is required.

Tony May
www.tonymayimages.com

[julius0377]

For example - shooting the opal lens cast calibration plate against an overcast sky and taking the grey balance in the centre - reading in a straight line across the centre of the frame (not in the corners) I get the following RGB values :  RH edge 174,171,170    Centre 176,176,176    LH edge 168, 171, 161.  The edge readings are near the edge of frame but not right at the edge.  This is the second back I have tried that has shown this problem.

I got a similar problem with my own and a loaner P45+ backs shooting straight with the H cameras. The cast was uniform from P-back to P-back and lens to lens (exactly the same for both backs and different lenses). So you could just use 1 really well made calibration shot, and be done with it.

Phase's answer is that its a feature of the chip and to use a Lens Cast Calibration on every single shot, which I find extremely surprising as all the marketing suggests this isn't needed unless using movements. And is only really any kind of answer if it can be guaranteed that only one calibration shot per lens is required.

This is the same answer they gave me after having my back in for a check.

Also shooting blown white backgrounds produces a lot of color noise with p+ backs compared to for instance Hasselblads own systems. The P+ backs exhibit color pixels and patterns in the 254/254/254 range. Phase told me that "this is outside of what a printer and screen will show and not a problem". This does not happen if the white background is not completely blown out. Both backs show this color noise, tiff files that show this are produced with C1.

[imagetone]

I got a similar problem with my own and a loaner P45+ backs shooting straight with the H cameras. The cast was uniform from P-back to P-back and lens to lens (exactly the same for both backs and different lenses). So you could just use 1 really well made calibration shot, and be done with it.

Thanks Julius, so are you still using your back and routinely applying an LCC on every single shot? You find this acceptable/workable?   Are the casts as strong as the ones I see?  You suggest that you don't need an LCC for each different lens and aperture. I am still testing to see if one LCC will work at different densities and apertures with the same lens but so far I don't see the LCC for one lens being appropriate for a different lens in my test files.

Also shooting blown white backgrounds produces a lot of color noise with p+ backs compared to for instance Hasselblads own systems. The P+ backs exhibit color pixels and patterns in the 254/254/254 range. Phase told me that "this is outside of what a printer and screen will show and not a problem". This does not happen if the white background is not completely blown out. Both backs show this color noise, tiff files that show this are produced with C1.

 

I haven't seen the colour noise with a P25+ in the studio using a backlit diffusing film background that is blown out - I just maybe get one point extra of red in the transition from pure white to some tone - that is, through 254 to 253 then 252 - but I presume this a white balance issue. I will recheck one of the TIFF files.  

Do you know if the equivalent Hasselblad backs give the colour casts? Are there corrections in Phocus for this and are they automatic?

Tony May
www.tonymayimages.com

[ThierryH]

This is the Kodak 22,2 MPx sensor which is very prone to these casts, with or WITHOUT shifts/tilts/swings, and with or WITHOUT short focal lengths lenses.

I can only recommend to make a white shading (like I wrote it in another thread a few days ago, here: http://luminous-landscape.com/forum/index....howtopic=34784) with a normal lens (80mm) at f8 or f11, and to use this as the sensor reference, instead of the one provided by the manufacturer: it should get rid of any cast when using the lens(es) straight (without shift/tilt/swing).

I hope this helps.

Best regards,
Thierry

I have just upgraded to a Phase One P25+ using it on an H1 with 80mm, 150mm and 35mm lenses. When using these lenses straight (without tilts or shifts) the files, without any adjustments, show quite marked colour casts with all three lenses.

For example - shooting the opal lens cast calibration plate against an overcast sky and taking the grey balance in the centre - reading in a straight line across the centre of the frame (not in the corners) I get the following RGB values :  RH edge 174,171,170    Centre 176,176,176    LH edge 168, 171, 161.  The edge readings are near the edge of frame but not right at the edge.  This is the second back I have tried that has shown this problem.

Has anyone else seen colour casts of this magnitude when using normal MF lenses with the Kodak sensors used in the P25+ and similar backs from Phase, Hasseldblad or others?  How have they dealt with it?

Phase's answer is that its a feature of the chip and to use a Lens Cast Calibration on every single shot, which I find extremely surprising as all the marketing suggests this isn't needed unless using movements. And is only really any kind of answer if it can be guaranteed that only one calibration shot per lens is required.

Tony May
www.tonymayimages.com

[clawery]

Thanks Julius, so are you still using your back and routinely applying an LCC on every single shot? You find this acceptable/workable?   Are the casts as strong as the ones I see?  You suggest that you don't need an LCC for each different lens and aperture. I am still testing to see if one LCC will work at different densities and apertures with the same lens but so far I don't see the LCC for one lens being appropriate for a different lens in my test files.

 
Tony May
www.tonymayimages.com

Tony,

You will need separate LCCs for each lens, but not at variable densities or apertures.  I would suggest that you use a standard aperture and capture a LCC for lenses that you see a lens cast.

[attachment=13857:LCC_pdf_1.jpg]
[attachment=13858:LCC_pdf_2.jpg]
[attachment=13859:LCC_pdf_3.jpg]
[attachment=13861:LCC_pdf_4.jpg]
[attachment=13862:LCC_pdf_5.jpg]

Chris Lawery(e-mail Me)
__________________
Sales Manager, Capture Integration
Phase One, Canon, Apple, Profoto, Eizo & More
National: 877.217.9870  |  Cell: 404.234.5195
Newsletter: Read Latest or Sign Up

[Guy Mancuso]

Very Interesting , never seemed to have them ever with the Mamiya lenses and the Phase body. This is with the Plus back. Hmmm .

[julius0377]

Very Interesting , never seemed to have them ever with the Mamiya lenses and the Phase body. This is with the Plus back. Hmmm .

They appear very faint (unseeable to my eyes). And can often be obscured by shadows cast from persons on the background and other elements making it difficult to pick up.

It would be interesting to see if a person owning a P+ back on a mamiya body (25 or 45) would capture a totally even white surface thats evenly lit and post the raw file. I'll see if I can do the same with my H-body and check for color casts plus post the raw, but I'm out of the studio for a couple of weeks...

[Guy Mancuso]

I would try it but just recently sold the P25+ for the P30+. Which seems to be working out nicely. The 6.8 micron sensor has a smoother look to it I am noticing unless I hit the clarity in C1 to a higher level.

[John_Black]

Tony - My P25 (non Plus) had cast on every image regardless of the lens used.  Very annoying.  And the dealer's & Phase's response was - the back is within spec, etc.  The RGB shifts where in the range +/- 3 when measured like you did.

I created a LCC profile for each lens in the studio illuminating a wall with two softboxes.  Two LCC profiles were made - one portrait, one landscape.  Each was shot F11 w/ ~ +1.5 stops of EC adjustment.  I wanted them hot, but nothing clipped.  Previously I had gone through very controlled testing at F2.8 through F16, shooting indoors, outdoors, sunny, overcast, etc.  In the end I found the F11 profile worked just as well.  Things get tricky if the profiles are done at wider apertures because light fall off plays a role too, so you're not just correcting cast, but also some fall off.

For the first couple months having to apply LCC's really ticked me off.  After awhile it just because part of the workflow and I got use to it, but it still ticked me off and eventually I sold the P25.  Before selling the back I looked into upgrading to a P25+ in hopes of resolving the cast issues.  When pushed, the dealer could provide no assurances that a P25+ would be any better.  

Later I bought a Mamiya ZD dSLR (Dalsa sensor) and the damn thing hasn't shown a hint of sensor cast in 3000+ frames.  And trust me, after the P25 I'm hyper-sensitive to sensor cast.  I'm not saying you should buy a ZD   But it does prove that there are backs out there that either don't have cast issues and/or the manufacturer stores a custom calibration file in the back's ROM to cancel out the base-line cast before we ever see it (I think Hasselblad does this).  In talking with other digital back owners (different brands), I came to the conclusion that the P25 was among the worst offenders when it comes to sensor cast.

[imagetone]

For the first couple months having to apply LCC's really ticked me off.  After awhile it just because part of the workflow and I got use to it, but it still ticked me off and eventually I sold the P25. ..........

.......... But it does prove that there are backs out there that either don't have cast issues and/or the manufacturer stores a custom calibration file in the back's ROM to cancel out the base-line cast before we ever see it (I think Hasselblad does this).  In talking with other digital back owners (different brands), I came to the conclusion that the P25 was among the worst offenders when it comes to sensor cast.

John, I'm very glad you posted this as your experience and feelings on this are very relevant to me.  I have been shooting a load of calibrations at different apertures for each lens to try and satisfy myself if it is possible to have a single calibration which works accurately for each lens. As you say the thought of having to correct every frame ticks me off too especially as C1 can't sort RAW files by focal length. To have to worry about specific calibrations for different apertures or densities is a non-starter for me and having seen the results of shooting white walled interiors with and without casts I don't want to use uncorrected files.  

The RGB shifts were in the range +/- 3 when measured like you did.
.......
 In the end I found the F11 profile worked just as well.  Things get tricky if the profiles are done at wider apertures because light fall off plays a role too, so you're not just correcting cast, but also some fall off.

The degree of cast I see seems greater than you report or other users suggest - how much difference do you see at densities around 170 to 180?  I was assuming that for the calibration too light an exposure would not be so useful.  I have been shooting test calibrations mainly against an overcast sky with EC adjustments of +1 to +2.3.  Do you know exactly how light fall off on the calibration frame affects the correction? And conversely if using a uniformly exposed calibration frame how the correction works when applied to images having variable density across the image (ie most images).  I would hope that the relative adjustments to RGB values are greater in the shadows and less in the highlights - can you confirm this with confidence (not putting you on the spot too much I hope as this is Phase's problem)

Tony

[imagetone]

I would hope that the relative adjustments to RGB values are greater in the shadows and less in the highlights

I'm not sure if I got this bit the right way round, but I hope you know what I'm getting at

[ThierryH]

Tony,

Why are you taking your head with this and not trying my advice. It doesn't need one shading for each lens and at each f-stop. Make one single default shading, with e.g. a 80mm lens, at f8 or f11, and use this for all lenses. I know this works with the 22 MPx sensor in Sinarbacks, so there is no reason it does not with other manufacturers' back.

IF it happens that you have shifts, tilts or swings, then it is another story, and you would THEN need a shading for EACH situation.

Best regards,
Thierry

John, I'm very glad you posted this as your experience and feelings on this are very relevant to me.  I have been shooting a load of calibrations at different apertures for each lens to try and satisfy myself if it is possible to have a single calibration which works accurately for each lens. As you say the thought of having to correct every frame ticks me off too especially as C1 can't sort RAW files by focal length. To have to worry about specific calibrations for different apertures or densities is a non-starter for me and having seen the results of shooting white walled interiors with and without casts I don't want to use uncorrected files.  

The degree of cast I see seems greater than you report or other users suggest - how much difference do you see at densities around 170 to 180?  I was assuming that for the calibration too light an exposure would not be so useful.  I have been shooting test calibrations mainly against an overcast sky with EC adjustments of +1 to +2.3.  Do you know exactly how light fall off on the calibration frame affects the correction? And conversely if using a uniformly exposed calibration frame how the correction works when applied to images having variable density across the image (ie most images).  I would hope that the relative adjustments to RGB values are greater in the shadows and less in the highlights - can you confirm this with confidence (not putting you on the spot too much I hope as this is Phase's problem)

Tony

[imagetone]

Tony,

Why are you taking your head with this and not trying my advice. It doesn't need one shading for each lens and at each f-stop. Make one single default shading, with e.g. a 80mm lens, at f8 or f11, and use this for all lenses.

Thierry, thanks for your advice, it is welcome.  I have been trying that of course. The cast on the 35mm is much stronger than the 80 and the 150 is in between, so how is a single adjustment going to work?  It doesn't so far work quite right every time using one adjustment per lens but its possible I am not getting some of my calibration tests lit uniformly.

From a theory point of view how does one uniformly lit calibration frame correct for different densities across the frame, light fall etc? I would like to be convinced by my eyes (not yet) and my head (not yet).

Tony

[ThierryH]

Tony,

Take an opal glas, thick enough to diffuse the light enough and evenly. You don't necessarily have to "lit" it with strobes or else. Make sure it is exposed correctly (about 2 stops more light than a normal well exposed capture through a lens). This will correct automatically the sensor's uneveness accross the whole surface.
One could think that the "sensor reference" delivered with each back and specific to each back/sensor does this work: but the practice shows that it does not, most of the time. The reason being, for Sinar at least, that this factory sensor reference is done without lens. The practice shows that this does not work well, and that any sensor reference made with a lens (any lens) will give better results concerning CC's.
Even if you are saying that your 35mm does exhibit more cast than the longer lenses, try it out, it will at least improve your results already by a fair amount of your shots, if not all.

Best regards,
Thierry

From a theory point of view how does one uniformly lit calibration frame correct for different densities across the frame, light fall etc? I would like to be convinced by my eyes (not yet) and my head (not yet).

Tony

[imagetone]

Thanks Thierry, my choice of words may have caused a misunderstanding. I wasn't asking how to make the reference shot although its good of you to spell it out and that is how I have been doing it.  The query was really how can  a single reference apply to every lens when the cast is different with each one.

I have made a correction for each lens, but they don't work perfectly and not every time. When you're not shooting white, neutral or uniform colour backgrounds its maybe not such a big deal. But it is unexpected with fixed lenses (I did expect it with tilts and shifts) and makes me wonder if I've spent my money on the right piece of kit.  If only I didn't love that large viewfinder and the money I invested in the H system.....anyway I have to go and do some real paid work now instead of spending time working out how I cope with the inadequacies of my expensive investment in old technology.

Tony

[ThierryH]

Tony,

Thanks Thierry, my choice of words may have caused a misunderstanding. I wasn't asking how to make the reference shot although its good of you to spell it out and that is how I have been doing it.  The query was really how can  a single reference apply to every lens when the cast is different with each one.

It can simply because your CC is probably not coming from the lens, but from the sensor ...

I have made a correction for each lens, but they don't work perfectly and not every time.
Tony

... which explains also why it does not work well, when doing for each lens.

Best regards,
Thierry

[imagetone]

It can simply because your CC is probably not coming from the lens, but from the sensor ...

If that's the casse why isn't the factory calibration sufficient to fix it?
Regards
Tony

[ThierryH]

That's my experience, with Sinarbacks, most probably because it is done WITHOUT a lens, with light directly on the sensor to create the sensor reference. The practice has shown (not only myself, but many other photographers) that a simple new reference file shot with a lens gives much better results.
I suspect that other manufacturers do their sensor references the same way, although I could be wrong.
It is however worth the test.

Best regards,
Thierry

If that's the casse why isn't the factory calibration sufficient to fix it?
Regards
Tony