This thread needs more pictures:
Back is P20
This thread needs more pictures:
Back is P20
2 Member(s) liked this post
All opinions are good and I've learnt a lot reading them all here.
1 Member(s) liked this post
At the risk of continuing down a road long travelled, can I ask a few questions, please?
- why not try one of the different 40mm lenses? Regardless of the MTF charts, it is likely there are better examples out there.
- given your rather neutral, or even not-so-happy experience with your MF camera setup, why so much interest in its performance?
- if there is a particular interest in seeing if it can be improved, might you consider Jack's suggestion of C1v9?
I'm a professional engineer and an amateur artist (some of my work can be seen here: http://linhof.com/en/portfolio/ander...ometric-chaos/ ), so approach camera gear from both technical and artistic aspects.
When it comes to P45+ and Capture One, what it makes better than say LR is 1) well-tuned noise reduction that increases apparent dynamic range, 2) demosaicing which is better at avoiding aliasing artifacts, and 3) well-designed color profiles, although heavily subjective.
Both Phase One and Leaf have chosen to make highly subjective color, while Hasselblad has more neutral rendition. It's fine, and it's a matter of taste what you prefer. I prefer Hasselblad's approach, but as I make my own color profiles it doesn't really matter to me which manufacturer I use. I wouldn't talk about "color fidelity" when it comes to Phase One, but rather "pleasing subjective color". Adobe's profiles aren't as well-designed generally, but many apply heavy post-processing in any case and find LR be perfectly workable.
I have personally not been able to spot any specific "medium format magic" that makes say the P45+ better than a Sony A7r just because it's medium format. Easy access to well-designed color profiles I'd say is the greatest advantage, but say if you like me is not particularly fond of Phase One's subjective look, you still need to make your own, or tune color to your liking in post-processing.
I'd also say that the well-tuned noise reduction is not necessarily an advantage; noise reduction hurts colors and make it look "digital" (pastel brownish). To me color is more important than being grain free so I actually use a third-party raw converter where noise reduction can be turned off completely. While I suffer a bit more aliasing in that as the demosaicer is not as well-tuned, the difference is marginal, especially as I shoot at smaller apertures and the images doesn't contain that bad aliasing to start with.
If my photography would be 100% about the result and nothing about the process, I would probably not use medium format today. Sure the IQ3 100MP is clearly better than anything else on the market, but it just ain't worth it to me as the best 135 cameras makes so good results today (properly handled -- lens selection and color profiles is more difficult to manage but I have the know-how to do it). I become very curious when some say that a legacy back like the P45+ makes way superior prints to say an A7r. If I had the possibility/time/patience I'd love to arrange a blind test. When I've worked with profiling I've noted that I'm better than many others to spot various aspects of color, but still I don't really think I would succeed in such a blind test. Therefore I remain skeptical about that others will, but I don't exclude the possibility.
It's the artistic side of me that makes me use medium format, and even considering large format film. I find it much more pleasing to shoot a P45+ on a Linhof Techno, than an overly electronic A7r, and it suits my artistic context. With sadness I see how key parts of the MFD industry seems to have forgot the process part of photography and just aims for a superior end result, even at a time when superiority is rapidly losing significance. I think medium format should be about flexibility and diversity, but the trend (slowly) goes towards making the same products as in the smaller formats but with bigger sensors with better numbers. On the positive side medium format is becoming more affordable, so affordable that it's much easier to motivate even if the quality difference in practical means is not that big. It shall be interesting to see where MFD stands in ten years, I hope my somewhat negative view on the flexibility/diversity will turn out wrong...
In the best of worlds we will still have a multitude of choices considering cameras, while backs are much less expensive which makes the quality-vs-price much less controversial. We'll see.
1 Member(s) thanked for this post2 Member(s) liked this post
- That 40 mm lens is the only one I have. That said Distagon 50/4 CF and Distagon 60/4 CF has similar, but less pronounced issues. Factory MTF pretty much tells the story (*).
- My interest in it's performance is related quite a bit of me being interested in optics and imaging science. That is not saying I am a master of either.
- I frequently try my images with Capture One v7/v8/v9...
But honestly, one of the reasons I look at it that much said about medium format conflicts with common sense, or is simply untrue. Just some ancient examples:
- MFD backs have 16 bits data. Fact is that the only Phase One back having more than 14 bits worth of data is the IQ3 100. Older backs, including the IQ3 50 actually store data in 14 bits. This was demonstrated by Anders Torger. Anyway, using 16 bits of data for 12-13 bits of DR is pure nonsense.
- Sixteen bit colour, see above. Photoshop is actually 15 bits, unless you use HDR mode.
- Six stops higher DR than DSLRs.
- All that talk about tonality. If Ctein can pull magnificient prints out of micro four thirds do we really need a 50x40 mm sensor to achieve great tonality? https://luminous-landscape.com/video...rsation-ctein/
- Better DR in higlights. That is pure nonsense. The digital sensors are always used in the linear part of their response range. MFDs are a bit biased towards underexposure. ISO is rated high, Capture One's default processing is very bright, unless linear curve is used. So the system tricks the user to underexposure to protect highlights. But that underexposure increases noise. But default noise reduction in C1 is quite hefty.
So offering another view may not hurt. That view or opinon is naturally also biased. I don't think there is an opinion without bias.
It is possible to measure things and correctly made measurements don't have bias, but they don't tell the whole story.
Here are bunch of sample images from the "Blad": http://echophoto.dnsalias.net/ekr/Ar...ernardSamples/
Some of those made as tests and others were just chosen to demonstrate a lens. Some folks bought into the Hassy system after checking those samples.
This generation Distagon 40/4 CF FLE
The new Distagon 40/4 CFE IF
Some prices, divide by 8.55 for USD, the last one is a 40/4 IF
Last edited by ErikKaffehr; 18th July 2016 at 15:13.
Portfolio: http://echophoto.smugmug.com1 Member(s) liked this post
No, one cannot generalize, and moreover, all profiles were improved in C1-7 and later. If one has a P45+ back, I'd say it's worthwhile to play around between flash-gray and daylight profiles (for a daylight image) and on a per-image basis to see which suits it best; for the others, I would reco starting with the appropriate profile.
*Also note that with advanced color editor, you can "fix" any color in the image from somewhat global to a very narrow band, and then save that as a new "user" profile to apply to other images. I had one dedicated to daylight skintones for the P45+, but never needed one for my P65+ or IQ backs as the factory profiles are excellent.
"Perfection is not attainable. But if we chase perfection, we can catch excellence."
Thanks Jack and Chris!
When I started with C1 v8 (now working on v9) for my IQ160 files, "flash" was set as the default profile. In futzing around a bit I don't see too much difference between this, "flash-easy grey" and "outdoor daylight," although I can see the histogram moving around a bit. Perhaps I am limited by my sRGB monitor (the files are in default color space, which as I understand is pretty close to ProPhoto RGB) or I'm just too inexperienced to see much difference. Perhaps both. I also see a "flash v2" profile, which does render a bit differently. If I only had any one of these to use I wouldn't complain, at least for outdoor work.
I created my own dual illuminant profile for ACR, which improved things a lot in terms of color but C1 still kicks the pants off of it in terms of final output. YMMV, of course.
Last edited by jng; 18th July 2016 at 07:42.
I have very few portrait shoots, but here is one that I like. This one was on the P45+. Processing on both ones is pretty much default.
Adjust exposure to taste
WB on sunlit pants on lady on the right
DCamProf dual illuminant profile
Adjust exposure to taste
WB on sunlit pants on lady on the right
Note: Lightroom has my sharpening preset that is intended for landscapes, C1 at default. LR left C1 right.
Arrgh! Resampling! Here are original images!
Last edited by ErikKaffehr; 18th July 2016 at 09:08.
Comments on Erik's DCamProf/LR vs C1 example, with reservation that I'm at an uncalibrated screen for the moment. In this light/setting I actually prefer the skintones from C1, it looks too yellow/saturated with DCamProf, I'm not fully pleased with the C1 result either, but I'm suspecting a somewhat problematic light condition due to the low sun, and possibly problematic makeup. I like the skintone on the left girl's hand which is without makeup and in shadow.
A default DCamProf profile doesn't try to do anything subjective, only correct for perceptual color changes due to the contrast curve (there's no standard to do so, so it's a model subjectively developed with my eyes and a few helping ones). It's intended effect is to make as realistic colors as possible, that is if they're unpleasing in real life they should look unpleasing in the photo . To make MFD-style pleasing skintones one need to make some subjective tuning of them.
I can clearly see that C1 is the right, due to the yellow cast of most colors (almost as if white balance is different, but neutrals are neutral), bringing out a warm tone which I assume many find pleasing but I just find it untrue and rather prefer a more realistic rendering; if I want yellow I make yellow. However, in my example of a subjective profile ("neutral+") I do add some warmth to yellows and greens to make a more pleasing effect in this type of light (and arguably more realistic as a single white balance cannot really reproduce the eye/brain response at the scene of warm sunlight mixed with cool sky-lit shadows), but less so than Phase One does it (and less global), and actually very similar to how I've seen Hasselblad do it.
C1 and DCamProf contrast curves are similar but not exactly the same, the DCamProf profile has lower contrast. A simple difference in contrast can affect skintones more than one think, as well as transition into whitepoint in the case of high key portraits (not this one). When going further than making a default DCamProf profile carefully considering contrast curve and possibly making a custom one is one key aspect. Often it's a good idea to mimic the contrast curve from the manufacturer.
Personally I think the C1 curve looks a bit too harsh, too much crushing as a starting point, but certainly not bad as a subjective end output for this image. C1 allows you to select different curves though and as part of the contrast is built into the profile itself (eg you can never get truly linear rendering) the color distortions when using a different curve is mild, they have surely primarily designed the profile for the default curve though.
We were using one of those "golden" reflectors to lighten up the shadows. So some of that yellow comes from that reflector. Ouch… :-(
I normally don't shoot portraits, but these was an outing with the camera club and these nice ladies so I got this one and a few others.
Processing was essentially minimal.
I may add that I will recalculate those DCamProf profiles. I did not do glare correction and used reference values for the old colour checker (pre 2014).
A few questions:
- My understanding is that you have some "looks", can those be downloaded?
- You have also shot your own home made target, which paper did you use to print it? I have some problems finding OBA free paper.
You are probably right on the makeup, the lady on the left was coming directly from a soccer match while the one on the right was "full production".
Last edited by ErikKaffehr; 18th July 2016 at 14:56.
You can download a generic subtle look from my camera profiling tutorial, described in this section how it's used:
Making a camera profile with DCamProf, the easy way
Direct link to the JSON file here:
Here's a list of what adjustment this look does:
- longer rolloff to white in skin-tone range to improve look of high-key portraits
- shorter rolloff to white in cyan-blue-magenta to improve color of skies
- slight warmup of midtones and highs in greens and yellows for making sunlit areas stand out more in landscapes
- slight saturation increase, skin-tones excluded as well as already highly saturated colors
- slight reduction of green component in oranges to get better separation between greens and reds (for landscapes)
- make neutrals more neutral (increases color "purity" and helps noise reduction)
- compress the gamut towards AdobeRGB
The adjustments are so subtle that at first glance you won't see any difference (it's supposed to be that way), but if you make a A/B swap they stand out quite clearly.
Note that as this look is intended to work on any profile it doesn't do any hue adjustments/corrections. The P45+ has a Kodak sensor and Kodak had quite "subjective" CFAs, unlike Dalsa or modern CMOS. This means that while a modern sensor makes pretty accurate color with only a matrix, a Kodak won't, meaning that DCamProf will struggle a bit with a bunch of non-linear corrections which then are relaxed to avoid poor gradients. This means that the actual hue results in the end is a bit less predictable with the Kodak, and it might need manual tuning to some extent.
I too have a Kodak sensor in my H4D-50, and I did do some manual adjustments. Excluding the dark skintone patch from a CC24 target may be beneficial, described here step 6, exclude bad patches, and then manually verify the skintone hue and adjust via a look operator using curves, described here adjusting hue
It's not easy to adjust skintone hue by just looking at one image with nothing to compare with, what I do is to compare with other well-known high-quality profiles for the same camera and preferably shoot a portrait under a stable condition so you have one in fresh memory, and then tune to taste.
The first images in this section shows how it may look:
Attached here, first Phocus, then DCamProf without adjustments, and then DCamProf with small subjective adjustment.
I shot a portrait of myself in outdoor overcast weather. A studio professional would shoot a studio portrait with a model with makeup of course. While Phocus skintones certainly weren't bad they were quite a bit more magenta than realistic and I preferred a more realistic tone. As I could see myself in the mirror at the time of shooting I know for a fact that DCamProf makes a more realistic rendering. It's a matter of taste though and there is no profile that can suit all people's tastes or work equally well for all types of skin or light conditions. I imagine that Hasselblad have tuned their skintones for studio flash and model makeup, rather than unshaved engineers standing around outdoors. An advantage of adding in a bit magenta like Hassy is that you pull away skin from green, in some light/skin combinations a more true look may actually make the skin look a bit green which looks less natural than skin with a little bit of magenta.
As you can see the subjective adjustment I did is very small, it's difficult to see without layering the other on top and make an A/B swap. So you really need to be a perfectionist to step in and make these types of adjustments. Of course you can make much larger adjustments too, but the farther you want to pull the look away from the neutral default the more cumbersome in becomes, so DCamProf is really for those that prefer a neutral/realistic rendering possibly with smaller adjustments on top.
Forgot to answer the second question about the home-made target. First I'd say when it comes to making a profile for general-purpose photography rather than reproduction of a specific subject in specific light it's hard to improve on a simple target like the Xrite/Macbeth CC24.
Using one extra target as I do in my tutorial, even a self-made is just as much about demonstrating DCamProf's flexibility when it comes to targets as making an actual improvement. A simulation of how much improvement you can get with this type of combination you can read about here:
Making a camera profile with DCamProf
A problem with homemade targets printed on inkjets is that while saturation is high the spectral spread is quite poor, despite the multiple colorants modern inkjets have. So you should surely combine it with a base target such as the CC24. Here's a plot that shows a 210 patch inkjet-printed semi-glosst target, note the limited spectral spread:
While a CC24 is more evenly spread:
I don't remember exactly what paper I used now, but I would guess that it was Canson Infinity Baryta Photographique or Hahnemühle Photo Rag Pearl, or Canson Infinity Platine Fibre Rag, all those are semi-glossy and OBA-free.
Having targets that contain high saturation colors is asking for problems, so do make one with just a CC24 first before experimenting with higher saturation. There are a number of reasons why that is problematic. First they are very difficult to shoot without large glare issues, I'd say a controlled studio setup is almost necessary, and not just any setup you really need to take all precautions you can to minimize glare. The other is that while a modern sensor is good at matching CC24 linearly (that's the sensitivity metamerism index in DxOMark), it breaks down with high saturation colors, and the LUT stretching can become pretty bad, one should preferably monitor it and adjust manually to avoid bad gradients or fighting too much against the sensor. This is particularly true for the Kodak, but also some modern sensors. The challenges are usually magnified in low temperature light, like StdA.
Then there's how raw converters expect profiles to behave. If raw converters where developed from scratch today when the computers are much more powerful than in the 1990s maybe they would be a bit more color scientific and themselves handle gamut compression dynamically depending on subject. However this is not the case and raw converters don't handle high saturation colors well, meaning that the profile must contain some gamut compression. Hasselblad's profiles contains pretty heavy gamut compression, not sure about Phase One as I haven't studied that particular aspect but I guess it's large as all the others.
You can compress gamut with DCamProf too. In the end what often happens is that the saturated target pulls colors out into higher saturation, and then they are compressed back in again with gamut compression. So I'm not really sure there is such a big gain. Some hue accuracy I suppose.
At some point I'll do some experiments with flower photography where ultra-high saturation is common and see what behavior I get, I think I would learn lots from that, but I haven't done that yet. Today I'm actually a bit uncertain how good DCamProf with CC24 (or any other target) is at matching hue of super-saturated flowers.
I have done some flower studies though, but then not about accuracy but about tonality. A problem with deep highly saturated colors is that our eye has poor sensitivity to gradients there so a flower easily looks just like flat saturated blob. The solution there is to work more with lightness to work out some tonality. This is sort of a built-in subjective adjustments that exists in DCamProf's defaults. This is perhaps most clearly seen in reds where Adobe's profiles generally doesn't perform that well due to their reliance on their RGB-HSV curve. C1 is better, but gets tonality by pushing reds into shades of orange (their yellow cast does have a stabilizing effect in this regard).
Here's an illustrative example of the tonality issue of saturated flowers demonstrated by an older version of DCamProf. Instead of just compressing tightly the gamut compression works with lightness too which makes the tonality/structure of the flowers stand out much better.
Note: the flower image has prophoto ICC attached, if your web browser is not color managed it will not display correctly.
I did not have any C1 example shot left here, but what it does is to make the tulips orange and works in red-orange shades which also is a way to make the tonality show, although hue-incorrect.
Last edited by torger; 19th July 2016 at 07:01.
Obviously taking the thread a little OT, but want to add some comments about C1 for thread posterity:
Under the color editor tool is a tab named "Skintone." What this is is an exceptionally fine-grained color editor dedicated to skintone -- you can tweak hue, sat, lightness and "uniformity" and set feather and gradient widths for neighboring colors. With it, one can easily de-hue the garish gold reflector from the above image; uniformity levels off colors and will smooth for example the not so close color-match in the makeup on the model's chin above. One can also save their own personal skintone sets based on preference, and use them as skintone WB droppers for future images.
Next, in the main editing menu you have a "Styles" tool. This tool is extremely powerful and has many built-in options, including numerous skintone choices -- an easy way to get to a good WB for any skin quickly; then fine-tune with the skintone editor tool. Of course all of these can be edited or stacked and then saved for later use.
PS: Color editor could easily "fix" the flowers above to any desired hue or sat amount...
"Perfection is not attainable. But if we chase perfection, we can catch excellence."1 Member(s) thanked for this post2 Member(s) liked this post
That is a very good question and I have spent a considerable time looking for an answer. Recently I looked for good images to compare processing in Capture One and Lightroom and went trough a lot of images. I have something like 4950 images with the P45+. It's use has been drastically reduced in the last two years:
Year Total P45+ 2013 7490 1674 2014 7505 1949 2015 7881 1013 2016 3489 187
So, the P45+ has lost share all the time. Last year the A7rII entered my equipment, offering a lot of new options. So, what was causing the decline of the P45+?
- A change in travel habbits in 2015, more travel by air and less by car.
- I sort of discovered a high failure rate with the P45+ while shooting in churches. There were two main factors.
- The limited range of focal lengths affect compositions negatively.
- Focusing in dark is hard, missing live view
- Lacking tilt and shift
- To much work and no great returns
Looking more at the images I have found the many of the images were good, but possibly a bit static and boring.
One aspect was that the MFD kit didn't make it to interesting places.
The main reason to shoot with MFD was that I had reserves for large prints. But, that advantage was eliminated by and large with the arrival of the A7rII.
A major factor with the A7rII that is offers tilt and shift with a lot of lenses, giving me new options.
I have also spent some time with some large prints 31"x47" (or so). The prints I looked were 50% crops, but still reasonably sized, as large as you get from a desktop printer. Looking at close the A7rII was obviously superior. But moving back to say something like 1 m viewing distance the prints were pretty close and there may be a small advantage for the MFD when looking at longer distance. Any difference was subtle.
So, I very clearly feel that my P45 is going into retirement.
The most important factors are probably:
- Lens choice, including zooms
- Tilt and shift with almost any lens on the A7rII
- Accurate focus using live view on the A7rII
The reason that the Hasselblad doesn't go to interesting places is mostly that it is the equipment left behind when there are limitations.
The A7rII is incredibly flexible, at least for my needs. I combine it with a HCam Master TSII that gives me 5-10 mm shift with the Canon 16-35, but more importantly offers tilt with most lenses. That T&S part is important for a lot of stuff I want to do. Tilt is more important than shift.
Last edited by ErikKaffehr; 20th July 2016 at 15:56.
Portfolio: http://echophoto.smugmug.com1 Member(s) thanked for this post
I know about the color editor in C1 and played a bit with it. It's a really good tool, and indeed the possibilities to tune colors to your taste is much better than many other raw converters. I like it much more than the sliders in LR. I also like that they've added "grading" wheels or what they call them in the latest version, haven't really tried that but I do like that C1 takes color tuning seriously and provides the user with many possibilities.
However, manually fine-tuning colors is not for everyone. It's one thing to pull a saturation slider, another to do really subtle adjustments to hues etc, so it's still important that the default colors are good. As far as I understand the popularity of C1 over LR as raw converter is mainly for the default rendering, not the tuning possibilities.
"Perfection is not attainable. But if we chase perfection, we can catch excellence."1 Member(s) liked this post
I wish I knew how to make it sing like I have done with LR and PS. My learning curve is step as I have only been working with it for maybe 2 months so far. New camera and new software make my life challenging right about now.
We can't put color/profiles on a single scale from bad to good, as there's lots of subjectivity involved in terms of look.
Making colors "sing" or images "pop" doesn't mean the same to us all :-)
Say if you know a LR/PS workflow and like it, why not continue with it if you get great results? Believe your eyes. If it looks good, it is good. Sure you can get a little extra in fine-tuned noise reduction (not at all as important with CMOS) and demosaicing (also decreasing in importance with better microlenses and smaller pixels) by using the native raw converter but it's marginal (easy to compare too if unsure), it's not worth dumping your workflow for that.
If you don't get colors to work due to "bad" profiles (=profiles with a look that you don't like) however, you should consider trying something else, or try making a custom profile.
Today I found a 503CW to mount my Credo on and give things a try for myself. It had a split prism acute matte D screen and PM45 viewfinder (which was quite good, but I way prefer the standard WLF.)
Wide open with a 150mm lens inside under dim light, focus was bang on straight away and this essentially got rid of any worries I had before regarding focus. Admittedly a 150mm should be pretty easy to focus, but essentially how I'd want to use such a camera it'd be more than okay. I'm used to manual focusing anyway.
Just wish I had the funds to go for it now... Alas, tax department calls.