gray balance before taking photos and GTI N8

[kenyee]

I've been getting more into photography lately and acquired a whibal white balance card (http://www.rawworkflow.com/products/whibal/index.html).

What's interesting is that the resulting picture doesn't look like the room according to my eyes. The autowhitebalance picture (394) looks more like what I see because of the yellowish ambient lighting in the room.

391 is a picture of the room after I used manual white balance and whibal (that's the GTI N8 on the projector wall from the Simple Grey thread on avsforum in case you're curious). GTI N8 is a certified "neutral munsell N8 gray" that is sold at photography shops like Adorama and B&H Photo, and since we're after the perfect gray screen, I decided to use that as the front projector screen for my Panny AX100.

393 is a picture of the room after I did manual white balance using the GTI N8. What's interesting is that this more reddish than the whibal calibrated image which seems to imply that N8 is not a perfect gray. I'm very impressed w/ how the whibal adjusted the white balance cleanly.

Any color experts care to comment on why GTI N8 can't be used to white balance a camera?

p.s., sorry about the blurriness of the photos...they were taken at 1/4sec because of the low lighting at night.
p.p.s., this means if you're taking photos of your projector screens and you have any ambient lighting, you should be doing a manual white balance to take an accurate picture of your screen (I think).

[mechman]

kenyee,

Welcome to the Shack!! It's great to have you here!!

mech

[wbassett]

Welcome back mech, and same Kenyee... great to see you.

This is a very interesting post. Maybe we can get the Calibration and Screen Shot standards thread going here that nobody was interested in elsewhere.

I'll have to get a white balance card myself. What I was doing was setting my internal light meter before shots, but this may be better to get a representation of exactly what I see. I usually take enough pictures to fill my memory card and then sort through them for the ones that look the best and closest to what I see on the screen. Most look nothing like what is up there.

Good topic!

[mechman]

wbassett wrote:

Welcome back mech, and same Kenyee... great to see you.

This is a very interesting post. Maybe we can get the Calibration and Screen Shot standards thread going here that nobody was interested in elsewhere.

I'll have to get a white balance card myself. What I was doing was setting my internal light meter before shots, but this may be better to get a representation of exactly what I see. I usually take enough pictures to fill my memory card and then sort through them for the ones that look the best and closest to what I see on the screen. Most look nothing like what is up there.

Good topic!

Thanks Bill! It was a very rejuvenating vacation!! Now I've got a lot to wade through!

I'll reply more on topic later (tomorrow).

mech

[mechman]

kenyee,

Can you label those photos? Is the top one 391? Thanks! Also I ordered a whibal card today. How long did it take till you received it? Kinda wish I would have had it before I went on vacation...

mech

[lcaillo]

What camera and how does it reference white? Could be abias in the camera or there may be other settings for the white point.

[kenyee]

mech: top one is 394, 391 is 2nd one, 393 is last (you can right click and select properties to see which file is which). I bought my WhiBal from B&H photo and it took less than a week. Let me know how you like it. Be careful mixing light sources...last weekend, I white balanced for interior lighting and took pics w/ the window in the background and the sunlight in the window was blue because the camera was adjusted for incandescent when I told it to white balance on the WhiBal so I had to crop out pics w/ windows in the background :-)

lcaillo: the camera is a Fuji F30 set on auto white balance. All cameras I've tried suck at auto white balance indoors because there are so many lighting temps that are possible (tungsten, incandescent, HMI, fluorescent, etc.). There are actually multiple fluorescent temps now as they compact fluorescent makers try to get closer to sun light temps.

[mechman]

The N8 and the whibal look fairly similar. Maybe a bit more of a green push in the whibal picture.

When I get mine in a week or so we'll know more.

mech

[kenyee]

The N8 picture definitely looks "warmer", so that makes me wonder if GTI did that deliberately (would make skin tones "glow" more). The N8 screen looks reddish in that picture (which is very odd because you'd think if I pointed at it and said "that's grey", it would be grey). The steam heat pipe in the corner is definitely reddish in that picture, but is painted with beige high-temp paint spray that you can buy from Home Depot. The beige paint on the wall is definitely beige and has no reddish tint to it in real life, even if it looks relatively warm in the N8 picture.

So for actual colors of the objects, the WhiBal one probably is more realistic. However, under the ambient light, everything does have a yellowish tone like the autowhitebalanced pic.

Of course, then I'm viewing this on an uncalibrated monitor (although it has been pretty accurate when I compare photos w/ what people say they took), so who knows? All this calibration stuff is pretty complicated :-)

[mechman]

In MSPaint the pipe gives me these RGB values:

In DSCF0391 171-164-148 Whibal
In DSCF0393 183-170-161 GTI N8
In DSCF0394 193-173-140 Auto

On auto mode there's quite a jump between the RGB values. Keep in mind that there's nothing scientific about this. I could have been off a few pixels (more than likely I was off many!) on each selection. With the N8 you still see a somewhat large drop between the Red and Green but very little between the Green and Blue. And the Whibal has an even smaller drop between Red and Green and then the big drop to Blue. You'd think there would have been a little more similarity there. I'm gonna write it off to poor pixel selection between the samples though. But I do plan on plugging away at this more when my Whibal card shows up. Should be here by Wednesday.

kenyee,

Did you get the GTI N8 paint from B&H as well?

[wbassett]

In case you are interested, here is the spectral values for GTI N8

Warren Schaefer wrote:

Hi Bill,
We are using an X-Rite Spectrophotometer, Illuminant D65.
The Spectral Values that we have recorded for our N8 equivalent neutral gray paint during January are:

L + 79.61
a - 0.93
b + 0.48

Let me know if you need any additional information.

Regards,

Warren Schaefer
GTI, Graphic Technology Inc.

As per the Lindbloom Calculator this comes out to an RGB value of 195.83 197.76 196.33 with a Yxy of 55.993290 0.312233 0.330739. EasyRGB comes out to a value of 196 198 196, which is obviously rounded from the Lindbloom figures and I consider both to be accurate.

Our target D65 neutral value is 0.3127 0.3290 , so it is close but falls slightly above D65 and left ever so slightly into green territory. This isn't a large move, but it isn't dead on like I expected it to be. It is still one of the nicer grays I have seen, albeit a bit more expensive than most people will want to pay. $25 isn't bad though for a pint and Kenyee seemed to get good distance out of it coverage wise.

Maybe these values help with what you guys are seeing.

Bill

[mechman]

wbassett wrote:

In case you are interested, here is the spectral values for GTI N8

As per the Lindbloom Calculator this comes out to an RGB value of 195.83 197.76 196.33 with a Yxy of 55.993290 0.312233 0.330739. EasyRGB comes out to a value of 196 198 196, which is obviously rounded from the Lindbloom figures and I consider both to be accurate.

Our target D65 neutral value is 0.3127 0.3290 , so it is close but falls slightly above D65 and left ever so slightly into green territory. This isn't a large move, but it isn't dead on like I expected it to be. It is still one of the nicer grays I have seen, albeit a bit more expensive than most people will want to pay. $25 isn't bad though for a pint and Kenyee seemed to get good distance out of it coverage wise.

Maybe these values help with what you guys are seeing.

Bill

You know Bill, you make me feel better and better each day! Now that I've finally read the DIY Painted Screens thread and most of the neutral gray thread it seems like my FG is pretty much on par colorwise with a neutral manufactured screen. For those of you reading this and wondering "huh?", follow along with me. Let me start off by saying that Bill has been at this DIY screen stuff for well over a year now and he's dotted most of his i's and t's. One of the things we're discussing in this thread is the calibration of a camera. And this is probably something that should be accomplished when shooting screen shots. Most pros do so now with a neutral gray reference card in their first shot or two. They then use this to set the color balance on all succeeding shots within Photoshop or whatever software they're using, using that reference shot with the neutral gray card in it. This takes the place of manual white balancing, which when using a standard piece of paper can give you awful results such as this:



That's one of a blue push! The lights in the hallway are incandescents - not fluorescents! This blue push is a result of the bleaching of normal printer paper, giving it the bluish push. They say you can use a pringles lid to calibrate white balance manually but Pringles lids have been changed to clear. I've tried tissue paper and coffee filters with the same results as above. But I'm really digressing here. So in the links above I'm told that professional manufactured screens generally have a red and blue push - they lack a bit of green to them. Well so does my Fashion Grey Laminate. It's basically a $2000+ screen for about $200! For comparison Fashion Grey's numbers are Red-195 Green-188 Blue-194.

Enough of my tangent. I'll be doing further testing this week when my Whibal card shows up. I'll have posts here and I intend to continue my comparison of Fashion Grey and Designer White then.

[kenyee]

mechman wrote:

Did you get the GTI N8 paint from B&H as well?

Yep. It was the quart size. I luckily got it to cover my 98" screen (projector uses 92" of it). It seems to cover pretty well, but reeeeeeeks worse than Kilz primer. It was pretty dumb of me to use it during the winter. When I first put it on, it wasn't bad, but as it dried, it really smelled...

[mechman]

Whibal showed up today. Don't know if I'll get to anything today but definitely tomorrow. Excuse the excitement but sometimes I feel like a kid in a candy store when I get to take screen shots!

mech

[JimP]

Hi boys..............can a pro photographer join into the discussion?

First of all, if you are looking for color accuracy in a photo viewed on your monitor, you get into a body of knowledge known as color management. This is basically a way of saying that you have to have your monitor calibrated to a standard so that everyone looking at it will see the same thing and not have variances due to how each individual monitor is calibrated. Then you have the flip side of this where you record the scene as neutral as possible or what I do, which is to shoot the scene using the camera raw format and then in photoshop CS2 use the tool in raw converter to white balance the scene. For this to work, you need a known neutral color in the scene such as either a known white or know neutral gray. From there, for my use, I usually look at the scene and ask myself if the picture looks too warm or cool relative to a reference "shirley" (which is a reference image of various typical scenes and what they look like). As another poster mentioned, mixed lighting will be a problem and you often have to make judgement calls as to which color balance is more attractive.

As to photographing an interior scene with consumer grade cameras, the best advice I could give you is to avoid auto white balance and set the camera as follows. If its a room lit with light bulbs, use the tungsten setting on the camera. If your photographing a projected image on the screen in a totally dark room, use the white balance setting for either a heavy overcast day or a clear day under a tree. These setting are usually pretty close to 6500K. Most digital cameras have one or the other.

As for using a gray card in the scene as a reference point, that would presume you have some way of making accurate adjustments based on it being there. Most consumer camera won't do that nor do most people want to spend the money for expensive software, so it would be better to just photograph it with using the correct color balance settings given the lighting conditions.

Any questions?

[wbassett]

This thread is getting to be very interesting.

JimP wrote:

As to photographing an interior scene with consumer grade cameras, the best advice I could give you is to avoid auto white balance

Interesting because elsewhere when I had a very similar thread going about camera setup and how to take accurate pictures of screen shots, the resident heavy that shows the most of these type of pictures only recommended auto everything.

I have a Fuji S3100, not the best camera out there by no means, but also not the worse. I guess for a consumer and taking typical pictures that people shoot, it's a good camera. I have been able to get screen shots that look okay, but none look as good as it does in person, or as good as other people's screenies.

My interest is to learn how to take the best (most accurate) pictures of a screen without any software manipulation. Some people have pulled their images into Photoshop and cleaned them up there, but that leads to a lot of debate on photo manipulation and making the picture look better than it does. Everyone does agree that screenies are eye candy and not an ideal way for people to be judging screen performance, but since the people viewing various screen methods can literally live a thousand miles apart, sometimes a picture is their only 'reference'. (which is why I also like screen specs and data on them)

I'm looking forward to seeing more in this thread and what ideas and tips come up!

[kenyee]

JimP wrote:

As for using a gray card in the scene as a reference point, that would presume you have some way of making accurate adjustments based on it being there. Most consumer camera won't do that...so it would be better to just photograph it with using the correct color balance settings given the lighting conditions.

"Given the lighting conditions" is the main variable we'd like to take out by using a gray card...specifically:
- fluorescents are different color temps now
- you have ambient light from wall reflections

Most consumer cameras that only output JPEG also do have a manual white balance mode which should work as long as you have a good white balance reference like the whibal.

I really should just pop for a Pulse colorimeter...it's been on my shopping list for a while just so I can figure out if my screen is off or not :-)

[JimP]

Here is a good example of what I'm talking about.

Take a look at the two squares marked "incandescent, Auto" and "Incandescent, Incandescent"

Even on a slightly better than average consumer grade camera, there's too much variance.
http://www.dpreview.com/reviews/fuji...000/page12.asp

[JimP]

kenyee wrote:

"Given the lighting conditions" is the main variable we'd like to take out by using a gray card...specifically:
- fluorescents are different color temps now
- you have ambient light from wall reflections

Most consumer cameras that only output JPEG also do have a manual white balance mode which should work as long as you have a good white balance reference like the whibal.

I really should just pop for a Pulse colorimeter...it's been on my shopping list for a while just so I can figure out if my screen is off or not :-)

Fluorescents are a special problem in that they have a discontinuous light distribution and come in many flavors. With that said, there is a specialized flourescent bulb that I use to use to judge color balance in prints from my color darkroom that were suppose to be much more accurate (not the double peak in the reds) and aimed at something like 5200 kelvin.

So I might be of some good to this thread, tell me exactly what you're trying to do and I'll see if I can help.

It sounds like you're trying to photograph on screen images in a dark room and the home theater room with just ambient lighting without a projected image on the screen. Am I close?

[wbassett]

Jim in my case/interest, I take screen shots with an image being projected. Setting can be anywhere from at night with no lights and only the projector running, and then I usually take some shots with the lights on to show performance with lights on and lights off.

I also do some shooting during the day when there is tons of sunlight pouring in the room.

[JimP]

wbassett,

For a projected image in a totally dark room, I would think that you'd want to use a camera setting that is closest to D65. Although D65 isn't the same as 6500 degrees kelvin, 6500 K is close enough. So given that, you'd want to use a manual camera setting that is closest to that. For cameras where you can adjust white balance numerically, 6500 works. But for those who don't and have symbols that represent incandescent, direct sun, shade,etc,....you'd want to use either cloudy bright or something that resembles shade under a tree.

Now, keep in mind that directors will use color tone as a creative tool. Just because I like clean whites doesn't mean that the director choose to tone a scene yellow for warmth or as in the matrix, cool.

When photographing a room with the projector at 6500K, and interior lights at 3200K or if its daylight, 5100 to 6500, you'll have to make a judgement call which element in the picture needs to look right. I would probably choose that the projected image be the most accurate.



When illuminating a room with daylight, things get more complex as day light varies depending on whether its mid day, late day, shady day, etc. You'll probably have to try several settings that are within a close range. from 5,100K to 6500K.

[kenyee]

JimP wrote:

It sounds like you're trying to photograph on screen images in a dark room and the home theater room with just ambient lighting without a projected image on the screen. Am I close?

The 1st thing was to check whether GTI N8 paint is truly a "neutral gray" as claimed. That's why I used the whibal card as a comparison because it's certified individually to be neutral gray at 185,185,185. If both are truly neutral, the images taken after a manual white balance should look the same, and whites should be whites, blacks should be black, etc...

The 2nd thing is, I'm trying to come up w/ a way to take pictures of a projected image so that results can be compared from one DIY screen to another.
I think ambient lighting is "coloring" the result you see on-screen, so doing a manual white balance is supposed to take out this side-effect so that we can compare images that we post. The dpreview article is also why I think you need to use something like whibal to set your color balance. Each camera defines "incandescent" differently, and each light source can be a different color. You just can't eyeball it.


ken

[JimP]

ken,

You're right on about the whibal card provided the card is illuminated with the exact same light as the screen...up to a point. Some materials reflect light differently, but what you're doing is as close as you can reasonable get it.

As for photographing a projected image with ambient lighting, tell me if we're talking daylight screaming through windows or room lighting at night and which kind of bulbs.

Maybe if you can provide a link I can tell better what we're talking about.

[wbassett]

kenyee wrote:

The 1st thing was to check whether GTI N8 paint is truly a "neutral gray" as claimed. That's why I used the whibal card as a comparison because it's certified individually to be neutral gray at 185,185,185. If both are truly neutral, the images taken after a manual white balance should look the same, and whites should be whites, blacks should be black, etc...

I can answer this one, it is close, but it isn't dead center D65 neutral. I posted the CIE L*ab data earlier as well as the Yxy and RGB values. It's closer than most of the grays being used for screens in DIY, but for something that is ISO standards I was a bit surprised it wasn't better, as in nailing neutral.

Here is a CIE graph if you're interested.

[kenyee]

JimP wrote:

As for photographing a projected image with ambient lighting, tell me if we're talking daylight screaming through windows or room lighting at night and which kind of bulbs.

That's the thing....most people won't have any idea what kind of light it is (I know mine are compact fluorescents), so that's why I mentioned whibal as a common ground that is relatively easy to use (set camera to manual white balance, put card in same lighting, click camera on gray part of card, then take photos of your projector screen). Most people aren't going to fiddle around w/ the color temperature setting of their camera (which you can't do in most P&S used for posting pics) until it looks right :-)

p.s., thanks wbassett...I know the GTI N8 isn't perfectly neutral gray and the photos I took at the start of this thread confirm this. It would be interesting if you could measure a whibal card though ;-) So now I'm wondering if I can take the whibal card, go to Home Depot or Sherwin Williams and ask them to mix up a can of paint that matches it...