CFL's and Daylight balanced film making me blue

[Brian Drake]

Hi-

I re did my photo strobes and old tungsten hot lamps to CFL's. They are a series of soft boxes with 5500K 45 watt bulbs. They all have a CRI of 85. I use these to shoot my fine art for reproduction. Many of the whites in the paintings have taken on a blue cast. Yellows turned way too green. I though I solved white balance a decade ago... now I feel like a newbie. I am led to believe it must be the CFL's. I don't have a color temperature meter, or Kelvin meter, but I did use my digital camera as a test. At 5100k it looked pretty good, when I set to around 4100k, the work looked perfect.

I did use Kodak E100vs
You can also see the blue cast in some order on the fuji FP100C45 instants-

Any thoughts on the right CFLs to use? Should I use a filter system? Should I ditch and go back to strobes? I like hot lamps for composition and I love the CFL's, for placing in and around delicate art work, as well as portraits. I would like to find a way to work with them. Your help and comments would be appreciated-

[Bob Salomon]

CRI for critical photography of the entire light (bulb + reflector + diffuser) must be 95 or higher for correct color. Look for a much higher CRI system. Not the CRI of the bulb by itself.

[Bob Salomon]

Since you are in Scottsdale you should go have a talk with Rod at Photomark. Bring the images with you.

[Brian Drake]

Rod had no real thoughts accept to switch to e100g.... Steven thought I should try a CC30M filter.

[Brian Drake]

Bob-
Regarding your CRI comment. I have read about this a few times and even spoke about it at Photomark with Steven and Rod... problem is NO ONE has told me where to find CFL systems with that high of a CRI. Do you have any ideas on this?

Thanks for your help in advance.

[Drew Wiley]

CFL's are still in their adolescence as far as color correction is concerned. It might be awhile yet till there's a serious selection of them. Too many outfits rushing to market just to get on the ground floor of energy-efficiency. It's difficult even to get
published spectrograms right now. I get horrible eyestrain from all CFL's, just like the ghoulish tube fluorescents of former days. But this will all gradually change.

[Bob Salomon]

Bob-
Regarding your CRI comment. I have read about this a few times and even spoke about it at Photomark with Steven and Rod... problem is NO ONE has told me where to find CFL systems with that high of a CRI. Do you have any ideas on this?

Thanks for your help in advance.

We have light boxes for viewing film or slides with higher CRIs and we have copy lighting with much higher CRIs but you would have to ask lighting manufacturers what they may have. Since we do have units with Dulux tubes for copying with a CRI of 98+ it is possible for others to have them as well. Just remember that it is the system that has the needed CRI not just the tube.

[Bruce Watson]

I re did my photo strobes and old tungsten hot lamps to CFL's. They are a series of soft boxes with 5500K 45 watt bulbs. They all have a CRI of 85. I use these to shoot my fine art for reproduction. Many of the whites in the paintings have taken on a blue cast. Yellows turned way too green. I though I solved white balance a decade ago... now I feel like a newbie.


The problem here is that you aren't using bulbs that conform to the D50 standard, or the D65 standard if that's your preference. It's not just a white balance problem. There's a lot more to light than just the average color temperature and the CRI. Both are important, but there's more to it -- check out the standards for more.

The problem with florescents is that the light you see is made up of just a few wave lengths. It sounds like the ones you are using are giving you just two wave lengths, which would account for the abysmal CRI rating. It would also account for the color shifting you are seeing on film. The painting after all is just reflecting the light you shine on it. If the yellow pigments only see a green or blue light, what you get back is a greenish yellow. What else could you get back?

What you need for the kind of work you do is full spectrum lighting. Most florescents have way too many gaps in the spectrum to be really useful for any type of photography, let alone critical art reproduction.

My advice is use lighting designed for the duty. IOW, go back to the lighting you were using before.

[B.S.Kumar]

All fluorescent lights have "spikes" in the color of light they emit. Accurate filtering is required to reduce this to a large extent, and then further tweak the colors in Photoshop. Minolta Color Meters and Sinar color correction filters can be found cheaply on eBay now. The Alzo HMI lights are used by some Betterlight users as a cheaper alternative to the Northlights. http://alzodigital.com/online_store/alzo_2000_repro_kit.htm They're not as cheap as CFLs, but significantly cheaper than the Northlights. You might also want to check Videssence and Kino-Flo lights.

Cheers,
Kumar

[Henry Ambrose]

As the last two posts mention, you are experiencing the result of "spikes" in the spectral distribution of the light emitted by your lamps. Do a search for the spectral distribution graphs of the particular lamp you are using. You can usually find these on the lamp manufacturer's website. These documents will contain the real answer about your particular lamp's output and will guide you in selecting another lamp that will satisfy your needs. Its not just the CRI, so don't stop there. Get the complete spectral distribution information.

To correct this I suggest you find a lamp that fits your fixtures that gives a smooth distribution of the visible spectrum that most closely resembles natural daylight.

Alternately you can use slight magenta filtration over the lens to improve things, at least to the point that you can correct in Photoshop. If you choose this route, buy a few different value filters and shoot a test of one subject with varying filtration. Please note that filtration is a work-around and won't solve the problem but will get you closer. It won't make up for the absence of a particular wavelength of light nor will it eliminate the overabundance of a particular wavelength.

I suppose you want a transparency as the final reproduction since you mentioned e100vs. That's the last choice of film I'd make. Kodak EPP or maybe Fuji Astia would be the two best choices available for "natural" color rendition.

[Bob Salomon]

As the last two posts mention, you are experiencing the result of "spikes" in the spectral distribution of the light emitted by your lamps. Do a search for the spectral distribution graphs of the particular lamp you are using. You can usually find these on the lamp manufacturer's website. These documents will contain the real answer about your particular lamp's output and will guide you in selecting another lamp that will satisfy your needs. Its not just the CRI, so don't stop there. Get the complete spectral distribution information.

To correct this I suggest you find a lamp that fits your fixtures that gives a smooth distribution of the visible spectrum that most closely resembles natural daylight.

Alternately you can use slight magenta filtration over the lens to improve things, at least to the point that you can correct in Photoshop. If you choose this route, buy a few different value filters and shoot a test of one subject with varying filtration. Please note that filtration is a work-around and won't solve the problem but will get you closer. It won't make up for the absence of a particular wavelength of light nor will it eliminate the overabundance of a particular wavelength.

I suppose you want a transparency as the final reproduction since you mentioned e100vs. That's the last choice of film I'd make. Kodak EPP or maybe Fuji Astia would be the two best choices available for "natural" color rendition.

This is partially true.

Better manufacturers even out the spikes by using special reflective surfaces behind the tube and, if necessary, specially formulated diffusors in front of the tube to reduce the spikes and add the missing colors. That is why it is important that you know the color temperature and the CRI of the system and not the lamp itself.

We have shipped hundreds, if not thousands, of copy light fixtures using high CRI daylight Dulux tubes to libraries, museums, universities, police, FBI, art institutes, etc. ll who use them for accurate color reproduction on film and digital.

And for many of these applications the cool burning under prolonged use is as important as the color temperature when copying many of the items they shoot as they are very delicate and prone to damage from the heat from other light sources.

[Henry Ambrose]

This is partially true.

Better manufacturers even out the spikes by using special reflective surfaces behind the tube and, if necessary, specially formulated diffusors in front of the tube to reduce the spikes and add the missing colors. That is why it is important that you know the color temperature and the CRI of the system and not the lamp itself.

We have shipped hundreds, if not thousands, of copy light fixtures using high CRI daylight Dulux tubes to libraries, museums, universities, police, FBI, art institutes, etc. ll who use them for accurate color reproduction on film and digital.

And for many of these applications the cool burning under prolonged use is as important as the color temperature when copying many of the items they shoot as they are very delicate and prone to damage from the heat from other light sources.

Nope, what I wrote is completely true and based on hard earned experience. I was building color corrected lighting units for digital scanning cameras before your sales organization even knew there was a need for such a thing.

Why not name the fixtures you sell so he can look at the possibility of buying some or deciding that they cost too much? Where are the spectral distribution graphs for your lights? They might be good, so show us - please.

The original poster has choices of filtering, buying new lamps or new fixtures and lamps. Whatever he chooses, depending on CRI rating alone won't get him there.

[Bob Salomon]

Nope, what I wrote is completely true and based on hard earned experience. I was building color corrected lighting units for digital scanning cameras before your sales organization even knew there was a need for such a thing.

Why not name the fixtures you sell so he can look at the possibility of buying some or deciding that they cost too much? Where are the spectral distribution graphs for your lights? They might be good, so show us - please.

The original poster has choices of filtering, buying new lamps or new fixtures and lamps. Whatever he chooses, depending on CRI rating alone won't get him there.

Kaiser makes the lights. You can check the Kaiser web site for info.

http://www.kaiser-fototechnik.com/en/produkte/2_1_sortiment.asp?w=3

http://www.kaiser-fototechnik.com/en/produkte/2_1_produktanzeige.asp?nr=3431


We also sold the GO LIghts and we have been selling high CRI daylight lighting for over the 25 years which is longer then digital scanning cameras have been around since we distributed the very first one commercially available - the Rollei ScanPack for the 6006/6008SRC cameras. It predated the first Leaf camera.

[CG]

The OP mentions CFLs with a CRI of 85. I believe I've seen CFLs with much higher Color Rendering Indexes. They seem to be available well into the 90s which presumably would translate into improved rendering.

Even then, test shoots and filtering to improve things, shooting again, till one has a fair set of results, would seem best.

C

[rvhalejr]

A friend of mine (former wedding photographer) told me that there seemed
to be something about the white material used to filter flash and modeling
lights that (for what ever reason) eliminates color drift.

Exhaustive testing and research after this discussion brought me to this
startling conclusion:

One or Two 400 watt CFLs filtered through a quality white umbrella will act
as near perfect CRI modeling light source(s).

This assumes that the CRI color target luminescence has been dialed in by placing
gels (side by side) in front of the light to enhance the temperature
profile as desired .,.

Combine this will carefully calibrated fill flash(s) of the identical temperature
will produce outstanding results.

Using Ambient light that is the same color temperature as the filtered modeling
and flash source(s) is ideal. One light stand, ambient daylight and one flash
seem to work best as to many sources will make the subject appear flat.

This assumes that you are using D50 or D65 light sources (not both)
and are using the appropriate filters (on camera and/or light sources) and
white balance (the DSLR as a simple color sensitive spot meter to check
the lighting setup before exposing that beautiful piece of 4x5 portra nc).

The best target for showing the slightest blue (or red) drift is the white wedding
dress material (use a doll or xmas tree angel, they do not complain and
work cheap).

With Daylight it is important to use a colorimeter to find out if it can be used in the
mix or not. If its in the 8500K region (or above) its almost useless except as a
heavily filtered sole source.

Time of day, year and weather can produce 5000-10000K daylight that is not an
ideal CRI light source. Thats why UV, Skylight, ND, Polarizers and gels are very
handy. :cool:

Modeling light CFLs followed by a combination of gels (side by side) filtered
through a white umbrella work surprisingly well with temperature matching fill flash(es).

Part of this research included using the highest (and most expensive) rated CRI
CFL made. It's spectrum was also vastly improved with the gel and white umbrella
diffuser technique as evidenced by measured target color image values .,.

With respect to LF, this technique will work for any film (C41, E6 or BW) but the
set of balancing filters and gels will be unique for each sub-type and will most
likely not be interchangeable.

The Temperature of CFLs and Strobes also can vary from one unit to the next and
will change over time (sigh) :(

[rvhalejr]

You probably already know this already but be sure to scan the film in
16bits (preferably wet, 3200 dpi and tiff).

I'm not sure about the latest photoshop (or scanner) software but picture
window pro is (IMHO) the preferred tool for fine tuning color drift.

Sorry about the previous long winded post (but it is a passion) .,.

All the Best,

Richard

[Jim Cole]

Part of this research included using the highest (and most expensive) rated CRI
CFL made. It's spectrum was also vastly improved with the gel and white umbrella
diffuser technique as evidenced by measured target color image values .,.
:(

Richard,

What CFL bulbs are you referring to in your post?

Thanks,
Jim

[rvhalejr]

Originally Posted by rvhalejr View Post
Part of this research included using the highest (and most expensive) rated CRI
CFL made. It's spectrum was also vastly improved with the gel and white umbrella
diffuser technique as evidenced by measured target color image values .,.


> What CFL bulbs are you referring to in your post?

I'm hesitant to name names for a lot of different reasons. There used to be a
gretag macbeth station (expensive but a great industrial tool) and I happen to
run across a replacement bulb or two (not all that expensive). Something about
full spectrum D50 or D65 and Seven or Nine phosphors if I recall (Its been a
long time).

In theory one could build a couple of arrays of these things to the impress the
curator staff but the roscolux spectural filters and white diffusers (Weston white
umbrella and/or Softboxes) really flatten out the spikes and can saturate colors
as needed.

To validate the setup (beyond targets like gm's) artist paints that are known
to be out of gamut (compared to anything else in the physical world) can be
used (since they are recognized trouble makers). Getting a blank canvas
and painting squares with these colors might be a good idea. Then build a
spectral gel set that equalizes out-of-gamut squares with those on the gm target.

I have to admit my modeling light setup looks "ghastly" by itself (lol).
When combined with a flash (of the same mired value) creates a digital image
(from a ambient daylight white balanced scanned film and/or matching dslr) will
be very precise. When it comes to color the instruments and tests win out
over my eyes ever time.

You can use almost ANY 5000K or 6500K CFL bulb set. It does not even have to
be full spectrum (I like the really big ones). However museums use to not allow
any flash because of UV issues and high luminance burst (the cumulative effect
over time being the same as exposure to direct sunlight). D50 used to be the
closest to the display lighting they use (no daylight) so it or something lower
for modeling lights would be ok as long as the color targets tell us we are golden
OR it matches a look (intentional drift) they want.

I get better results with D65 on paintings as compared to D50 because there always
seems to be some (green building) indirect sunlight around. Then again morning and
evening usually require D50. Who knows, some galleries may use a combination
of yellow (D50) and/or blue (D65) to accentuate the work, and it may vary from
piece to piece (they pay the invoice so .,.).

The calibrated colorimetric standard used are color targets and ambient
daylight temperature values recorded through out the day with 120/220 film
and/or dslr and white balanced mired filters that yield the most closely
matched published values for the color target(s). A gray card, black and white
cloth targets are also used.

Sometimes there are compromises that have to made, but as long as absolute and
relative colorimetric values are well with the 16bit digital color editing working range
of our software tools the setup should all be good to go.

Then we pull out the 4x5 film (same as the 120/220 used in test shots).

[rvhalejr]

Sorry, that last reply was in response to Jim's Question.

One more thing (I'm agnostic about film and digital, both having strong and
weak points), I've seen a transparency that had no grain (way above 3200dpi)
and a digital image that exceeded the nyquist frequency[1] limit by a factor of 2.

These are extremely rare occurrences but do exist. Almost all claims to be able
to achieve these results on a regular basis should be dismissed out of hand (sigh).

That aside, we need to be aware of is the weakness of dslr's and digital backs
since we are using these for calibration tools and are a part of our (or at least
my) daily work flow.

Because of the well known RGBG arrangement slight differences in colors are
the Achilles Heal of the 40Mega-pixel chip.

The full color pixel value is 40mp/RGBG (four photo sites) == 10 megapixel color camera

I've been using and selling Charles Sleicher targets (RGB and Black on White)
for awhile. To see the 10 mega pixel effect make red line pair targets
on a red background with a sinusoidally decreasingly small difference in color.
At some point the scanned film image will show up to 2-3 times the resolution
(lp/mm) as compared to the dslr or digital back image (given 2G photosites and
perfect point algorithm mentioned in Note 1 on the bottom of a highly intelligent
and no doubt expensive, stack of software).

This is only being pointed out to show that the digital image will lose some
of its color acutance as it approaches the megapixel/RGBG limit.

Given what I've seen (both good and bad) from film and digital sensors I
would be comfortable estimating that a well exposed color balanced piece of
6x6 or greater mirror like flat piece on film (using the fine focusing loupe lp/mm
kit I sell) behind a good Schneider, Nikon, Zeiss etc. APO lens will out capture
the finest color detail of the large format 40 Mega pixel back (based on a
2800dpi 16bit nikon fluid scanning limit for color film).

I did not account for the less than 16bit per channel color depth of the kodak
KAF-39000-AAA chip (has anj internal 12bit dynamic range but outputs 16) so
just think of that as my four bit margin of safety.

A lot of 4x5 film verses digit test reviews have been popping up which I'm having
trouble with.

My Engineering Test Requirements
The first is I want to see a picture of the reflection of an unexposed
piece of film in the film holder to check for flatness. The reflected image
should be perfectly mirror like, demonstrating no out of the box film curl.
(As far as I know flatness can only be obtained by a specific unexposed
film taping technique in the black bag or dark room).

The second is I want to know is the reviewer used my fine focusing loupe kit (for
precision plenum photography) that will bring the lens image and emulsion plane
into perfect focus simultaneously (across the entire frame inspected on the
side and corner stand). Once observed (validates most of the camera system
and) it guarantees the best possible image resides within the approximately
.001 to .002" emulsion plane.

The reviewers will only need to pay actual postage to them AND back to me
(unless they want to buy it or craft their own from almost 70 pages of
documentation, photographs and drawings that comes with every kit).

I don't like to name names but some of great renown have done that and more
for me, so its time to give back to those who share one of my life's great
passions.

Note [1]
40Mega-pixel the monochromatic value. There also is a airy-disc (perfect-point)
algorithm being used that can make digital pictures "scary". But I digress, I do not
even know if its published (a tightly held trade secret no doubt). Simply put it is
possible (with high speed sampling deterministic algorithms) to achieve reliable sub
pixel accuracy (yes, exceeding the monochromatic 2d nyquist frequency limit in
something resembling a virtual 3d working space).

[rvhalejr]

Part of this research included using the highest (and most expensive) rated CRI
CFL made. It's spectrum was also vastly improved with the gel and white umbrella
diffuser technique as evidenced by measured target color image values .,.


> What CFL bulbs are you referring to in your post?

I'm hesitant to name names for a lot of different reasons. There used to be a
gretag macbeth station (expensive but a great industrial tool) and I happen to
run across a replacement bulb or two (not all that expensive). Something about
full spectrum D50 or D65 and Seven or Nine phosphors if I recall (Its been a
long time).

In theory one could build a couple of arrays of these things to the impress the
curator staff but the roscolux spectural filters and white diffusers (Weston white
umbrella and/or Softboxes) really flatten out the spikes and can saturate colors
as needed.

To validate the setup (beyond targets like gm's) artist paints that are known
to be out of gamut (compared to anything else in the physical world) can be
used (since they are recognized trouble makers). Getting a blank canvas
and painting squares with these colors might be a good idea. Then build a
spectral gel set that equalizes out-of-gamut squares with those on the gm target.

I have to admit my modeling light setup looks "ghastly" by itself (lol).
When combined with a flash (of the same mired value) creates a digital image
(from a ambient daylight white balanced scanned film and/or matching dslr) will
be very precise. When it comes to color the instruments and tests win out
over my eyes ever time.

You can use almost ANY 5000K or 6500K CFL bulb set. It does not even have to
be full spectrum (I like the really big ones). However museums use to not allow
any flash because of UV issues and high luminance burst (the cumulative effect
over time being the same as exposure to direct sunlight). D50 used to be the
closest to the display lighting they use (no daylight) so it or something lower
for modeling lights would be ok as long as the color targets tell us we are golden
OR it matches a look (intentional drift) they want.

I get better results with D65 on paintings as compared to D50 because there always
seems to be some (green building) indirect sunlight around. Then again morning and
evening usually require D50. Who knows, some galleries may use a combination
of yellow (D50) and/or blue (D65) to accentuate the work, and it may vary from
piece to piece (they pay the invoice so .,.).

The calibrated colorimetric standard used are color targets and ambient
daylight temperature values recorded through out the day with 120/220 film
and/or dslr and white balanced mired filters that yield the most closely
matched published values for the color target(s). A gray card, black and white
cloth targets are also used.

Sometimes there are compromises that have to made, but as long as absolute and
relative colorimetric values are well with the 16bit digital color editing working range
of our software tools the setup should all be good to go.

Then we pull out the 4x5 film (same as the 120/220 used in test shots).

[rvhalejr]

One more thing (I'm agnostic about film and digital, both having strong and
weak points), I've seen a transparency that had no grain (way above 3200dpi)
and a digital image that exceeded the nyquist frequency[1] limit by a factor of 2.

These are extremely rare occurrences but do exist. Almost all claims to be able
to achieve these results on a regular basis should be dismissed out of hand (sigh).

That aside, we need to be aware of is the weakness of dslr's and digital backs
since we are using these for calibration tools and are a part of our (or at least
my) daily work flow.

Because of the well known RGBG arrangement slight differences in colors are
the Achilles Heal of the 40Mega-pixel chip.

The full color pixel value is 40mp/RGBG (four photo sites) == 10 megapixel color camera

I've been using and selling Charles Sleicher targets (RGB and Black on White)
for awhile. To see the 10 mega pixel effect make red line pair targets
on a red background with a sinusoidally decreasingly small difference in color.
At some point the scanned film image will show up to 2-3 times the resolution
(lp/mm) as compared to the dslr or digital back image (given 2G photosites and
perfect point algorithm mentioned in Note 1 on the bottom of a highly intelligent
and no doubt expensive, stack of software).

This is only being pointed out to show that the digital image will lose some
of its color acutance as it approaches the megapixel/RGBG limit.

Given what I've seen (both good and bad) from film and digital sensors I
would be comfortable estimating that a well exposed color balanced piece of
6x6 or greater mirror like flat piece on film (using the fine focusing loupe lp/mm
kit I sell) behind a good Schneider, Nikon, Zeiss etc. APO lens will out capture
the finest color detail of the large format 40 Mega pixel back (based on a
2800dpi 16bit nikon fluid scanning limit for color film).

I did not account for the less than 16bit per channel color depth of the kodak
KAF-39000-AAA chip (has anj internal 12bit dynamic range but outputs 16) so
just think of that as my four bit margin of safety.

A lot of 4x5 film verses digit test reviews have been popping up which I'm having
trouble with.

My Engineering Test Requirements
The first is I want to see a picture of the reflection of an unexposed
piece of film in the film holder to check for flatness. The reflected image
should be perfectly mirror like, demonstrating no out of the box film curl.
(As far as I know flatness can only be obtained by a specific unexposed
film taping technique in the black bag or dark room).

The second is I want to know is the reviewer used my fine focusing loupe kit (for
precision plenum photography) that will bring the lens image and emulsion plane
into perfect focus simultaneously (across the entire frame inspected on the
side and corner stand). Once observed (validates most of the camera system
and) it guarantees the best possible image resides within the approximately
.001 to .002" emulsion plane.

The reviewers will only need to pay actual postage to them AND back to me
(unless they want to buy it or craft their own from almost 70 pages of
documentation, photographs and drawings that comes with every kit).

I don't like to name names but some of great renown have done that and more
for me, so its time to give back to those who share one of my life's great
passions.

Note [1]
40Mega-pixel the monochromatic value. There also is a airy-disc (perfect-point)
algorithm being used that can make digital pictures "scary". But I digress, I do not
even know if its published (a tightly held trade secret no doubt). Simply put it is
possible (with high speed sampling deterministic algorithms) to achieve reliable sub
pixel accuracy (yes, exceeding the monochromatic 2d nyquist frequency limit in
something resembling a virtual 3d working space).

There was also a very fine diffusion filter used in an attempt to emulate the
Super-Angulon 8/65mm (magic glass)

[rvhalejr]

You probably already know this already but be sure to scan the film in
16bits (preferably wet, 3200 dpi and tiff).

I'm not sure about the latest photoshop (or scanner) software but picture
window pro is (IMHO) the preferred tool for fine tuning color drift.

Sorry about the previous long winded post (but it is a passion) .,.

All the Best,

Richard

[Brian Drake]

As the original poster, I wanted to follow up. I appreciate every ones comments and thoughts. I did a digital shoot under my new CFL kit and it was great. I also did some black and white 35MM and they were great as well. So... I decided to keep the CFL's for certain digital work and went back to 3200k photofloods with blue filters and my 5500K strobes.(No i Don't use both at the same time) Guess what? Both look fantastic. My color problem was one of my own making. Thanks again to every ones comments.

[B.S.Kumar]

Which CFL kit did you buy?

Kumar

[Edwin Beckenbach]

One more thing (I'm agnostic about film and digital, both having strong and
weak points), I've seen a transparency that had no grain (way above 3200dpi)
and a digital image that exceeded the nyquist frequency[1] limit by a factor of 2.



I saw a troupe of six fingered gnomes playing dodge-ball with a Fabergé egg in my garden this morning. They said something about being the point spread function of a moonbeam.

[rvhalejr]

Should I ditch and go back to strobes? I like hot lamps for composition
and I love the CFL's, for placing in and around delicate art work, as well
as portraits.

Assuming you did not find a solution already, I always opt for a DIY solution
first before recommending the really wonderful several thousand dollar solution.

Working to a setup thats known to be good (perfect white on your film - see
notes below on film types) is a good way to sanity check your colorimeter
and/or dslr white balance.

If you do not use the CFL with a Softbox diffuser or white umbrella without
the black backing then go no further since light transmitted through
such a white medium tends to attenuates spikes by re-transmitting photons
that are statistically more likely to be equal parts of RGB (white light).

I consider it a calibration exercise, science project that re-affirms
theory on verifies re-reproducibility.

Photographers generally agree that its less time consuming to shoot
something right than muck around photoshop, picture window pro and/or
other types of software.

You use a STROBE (like most tend to) using daylight balance film
(D50 or D65) and when you shoot a white target paint, cloth,
paper, etc. you get a perfect white (RGB balance in 16 bits).

Balanced film (and strobe) around ambient daylight is exactly the same
temperature (otherwise blue tint would be seen since > 8500K
natural daylight tends to swamp out skylight filters).


Now we have all variables under control, except for the CFL.

Do you have a Roscolux Swatch Book (they used to be free)? For blue
balance look at straw, bastard amber, etc. and get an few test sheets
(only a few dollars).

Cut test sheets in 1/4" strips of a sheet and start taping them on the
CFL (with 1/4" gaps) and shooting test shots (keep the CFL on and let
it warm up for at least 15 minutes). I prefer to calibrated a single bulb
at a time. As strips are added to the light the blue tint will disappear.
If two many strips are added you will notice a yellow or amber tint
creeping into the white target.

I've used the same technique on my nikon stobes, the strips are more
like 1/8 - 1/16" wide over the bulb , it dials in so well that the white
diffuser is not need.

DIY is crude and time consuming. But without filters and diffusers even
the best CFL Full Spectrum Lighting system calibrated with the Minolta
color meter III (or similar device) will have some peaks and valleys
throughout the spectrum.

If there are several out-of-gamut artist colors to die for adjustment
of your lighting will be needed (unless the technology is much further
along now and they actually test for those paints, white wedding dresses,
material, black tux seams, and every swatch on your color target).

The only way to escape DIY is to have it serviced periodically to keep
it in calibration, as well as your strobe(s), or just replace everything
periodically.


Other Issues

Our worst enemy is winter daylight as it can tend towards 10,000K
(really, really bluer). :(

Velvia (and 100Vs) are almost impossible to balance because of the
intentional over-saturated look.

In the case of fuji any other E6 would be better, likewise Kodak 100G
and maybe 100GS should be an improvement.

C41 is usually the choice for ultra-fine color accuracy (at the cost of
more grain), I'm a portra addict .,.

Another variable (out of scope here is scanner calibration,) using targets
periodically or rescanning film known to be perfect is a good check to
make sure nothing has changed .,.

Out-of-gamut that kill me are all those colors out printers and monitors
cannot recreate.

Film and Digital Out-of-gamut colors seem to be less of an issue. I'm going
to try to ignore RGB, CYMK, CIE details for now. But if today's compressed
capture color space contains a 16bit true color image is accurate relative
colormetricly then it should be possible to map to a future expanded color
space (if need be) to drive color display devices that can accurately render
todays Out-of-gamut colors.

This last paragraph of verbosity is from an old film scanner soul getting up
the nerve to ask a poster in this thread if (someday) it will be possible (if
standards exist) to take out-of-gamut artist paints (maybe six
rgb, cym cie vectors) and add them into the tanks of a really big awesome
(sixteen reservoir ?) continuous ink system.

Then out-of-gamut would be gone (sort of in the case of hardcopy) .,.

[rvhalejr]

> Originally Posted by rvhalejr View Post
> One more thing (I'm agnostic about film and digital, both having strong and
> weak points), I've seen a transparency that had no grain (way above 3200dpi)
> and a digital image that exceeded the nyquist frequency[1] limit by a factor of 2.



I saw a troupe of six fingered gnomes playing dodge-ball with a Fabergé egg in my garden this morning. They said something about being the point spread function of a moonbeam.

So they were qualified radiometry experts discussing moon beam scatter and spectral emission line shift ;)

One out of 10,000 slides, its the holidays ok?

I've never seen a transparency (before or since) that would benefit from a
greater than 5000 ppi fluid scan, except for that one. It was old ISO 25
E6 Kodak film (legendary noise floor 1/2 that of ISO 50).

Maybe Nyquist sampling of sequential data got those gnomes (they might
have been department chairs at prestigious institutions).

If the data context is from an image, (not just a simple string of
bits) the matrix will have a huge amount of information that (in theory) can
be exploited to build a 3d model, effectively appearing to be subpixel (or emulsion
cloud) sampling, as an admittedly overly simplistic analogy, hence the
"factor of 2" comment.

Those gnomes do need to allow for the possibility (how ever remote) that
1 out of 10,000 sub N frequency sampled photos will not exhibit aliasing due
to some factor not often encountered or observed.

Is this worthy of publication and peer review ? No. Will there never be a
generally accepted dissertation on MTF and how it might be cracked,
like an egg ?

That thesis was originally outlined by 10,000 monkeys in a room jumping
up and down on keyboards.
------------------
I wanted to edit that post but then ran into the 120 minute time out (after the
wife started screaming in one ear "GO GET MORE RIBBIONS AND BOWS !!!").

[Edwin Beckenbach]

> Originally Posted by rvhalejr View Post
> One more thing (I'm agnostic about film and digital, both having strong and
> weak points), I've seen a transparency that had no grain (way above 3200dpi)
> and a digital image that exceeded the nyquist frequency[1] limit by a factor of 2.




So they were qualified radiometry experts discussing moon beam scatter and spectral emission line shift ;)

One out of 10,000 slides, its the holidays ok?

I've never seen a transparency (before or since) that would benefit from a
greater than 5000 ppi fluid scan, except for that one. It was old ISO 25
E6 Kodak film (legendary noise floor 1/2 that of ISO 50).

Maybe Nyquist sampling of sequential data got those gnomes (they might
have been department chairs at prestigious institutions).

If the data context is from an image, (not just a simple string of
bits) the matrix will have a huge amount of information that (in theory) can
be exploited to build a 3d model, effectively appearing to be subpixel (or emulsion
cloud) sampling, as an admittedly overly simplistic analogy, hence the
"factor of 2" comment.

Those gnomes do need to allow for the possibility (how ever remote) that
1 out of 10,000 sub N frequency sampled photos will not exhibit aliasing due
to some factor not often encountered or observed.

Is this worthy of publication and peer review ? No. Will there never be a
generally accepted dissertation on MTF and how it might be cracked,
like an egg ?

That thesis was originally outlined by 10,000 monkeys in a room jumping
up and down on keyboards.
------------------
I wanted to edit that post but then ran into the 120 minute time out (after the
wife started screaming in one ear "GO GET MORE RIBBIONS AND BOWS !!!").

High five on the very entertaining reply!

RAP100F holds a bit of detail above 2500cpi and so could benefit slightly from sampling a bit beyond 5000ppi but of course this would be with contrast that is far less than inspiring and of probably of marginal utility (not to mention beyond my ability to pull off in the real world). If there was truly a film in the distant past that exceeded this I would enjoy hearing more about it.

There is no unfiltered sub N frequency image that will be immune to aliasing in the real world. You can theoretically model what would be required but it won't look like anything we commonly think of as a photograph. However, this is the irony of the situation, if there is no aliasing there is also no contribution of higher frequencies in the sample to the frequency content which is resolved and if there is aliasing the contributed information is absolutely ambiguous. Also, if a given sample doesn't have sufficient information to represent subpixel frequencies in 2D how on earth do you presume it has enough to build a 3D model with perhaps orders of magnitude more degrees of freedom?

If you can do this it would be more than worthy of peer reviewed publication because it would be absolutely astounding to see a method that allows resolution beyond the folding frequency of the sampling system without some variation of oversampling (and statistical oversampling can be ruled out because we're only talking about one image) in which case the 'assumed' Nyquist frequency really isn't.

[rvhalejr]

High five on the very entertaining reply!
...
There is no unfiltered sub N frequency image that will be immune to aliasing in the real world.
...


I humbly submit that you are correct.

Spectral Transmission and CIE Chromaticity (from rosco.com/us/filters/roscolux.asp)
suggest two gels worthy of consideration in this threads context:

#06 No Color Straw (ideal suppression curve for blue tint in white)
#3317 Tough 1/8 Plusgreen (partial green to balance with fluorescents)

When beloved photographers become concerned about the ghastly glow
of wide spectrum filtered and diffused florescent lighting one can say
"Its Moonlight" and perhaps let out a little wolf howl at an imaginary full moon.

During the pursuit of a disciplines advancement conventional wisdom,
canonical facts and empirical truths might be considered temporary final
approximations in the moment.

[rvhalejr]

...
Other Issues

Our worst enemy is winter daylight as it can tend towards 10,000K
(really, really bluer). :(



See Attached Image, You can imagine what the 10K curve looks like.

Ref ...xritephoto.com/html/ECM%20Tech%20Brief-en.pdf

[rvhalejr]

Assuming you did not find a solution already, I always opt for a DIY solution first before recommending the really wonderful several thousand dollar solution.

Cokin P (Film + Most 4x5 lenses in a pinch)
FLW (P036) corrects the lighting from florescent "white" bulbs.
FLW (P138) is the graduated version for mixed lighting.
FLD (P046) corrects the lighting from florescent "daylight" bulbs.
FLD (P139) is the graduated version for mixed lighting.

Basic florescent filters :p