best way to calibrate production monitor to 5D?

[Josh Bass]

Since a DSLR doesn't generate color bars, what's the best way to calibrate a field monitor to the camera? Some suggest taking a still of NTSC bars and putting it on a CF card, then playing that out of the cam into a monitor and calibrating to that. Is this valid?

[Denis OKeefe]

Personally I don't think it is possible to really set up the 5D lcd with great accuracy, at best you get it in the ball park so what you see is pretty close to what you get. Then leave it alone.
A generated ( or played back off the CF card) bars chart is useful for setting a monitor but it doesn't tell you anything about what the camera is seeing, for that you need to shoot an actual bars/chip chart.
If you set up your monitor from a known source such as generated bars on your video camera, then without touching the monitor settings shoot bars or another color chart with the 5D you could first adjust lcd brightness so the two monitors look as close as possible, then adjust camera settings to get the monitor image to match what the chart looks like to your eye ( and waveform monitor if you have one).
Cameras, monitors, eyes, lenses and lights are all variables that are constantly in play; there is no "perfect" setting. The best you can achieve (in my opinion anyway) is a setting that you know to be consistent, that you understand is an approximation (hopefully close) of what the camera sees. Then have some faith in the gear and settings you are using.
Bear in mind that generated bars are "illuminated" by the lcd, led or crt screen of a monitor. A chart in front of the camera is reflecting light, so it will always look different than "camera bars".

[Josh Bass]

Did not realize it was such a bugaboo.

I have an XL2. . .if I run bars from that to the monitor, calibrate, is that better than nothing? Even though I'm not using the XL2 at all, at least I'm setting up to NTSC color bars right?

Don't have a chip chart/etc. for the other part.

I guess it'll be funny in a few years when we don't have to deal with this crap 'cause we're all using AF100s and F23s and whatnot.

[Denis OKeefe]

Hello Josh -
That will set your monitor to a known standard, reducing at least one of the variables. Then, even without a chart you can adjust the Canon picture settings to something approximating reality on that "set" monitor.
Ideally you are trying to get the camera image on the monitor to look like what you eye is seeing.
At the very least there are three variables - the camera, the camera LCD, and the monitor. You have to set one and then leave it alone as the standard, then adjust the others to match.
Set your monitor from known bars on the video camera, then set the camera to match video "reality" (such as a non moving bookcase or static outdoor shot). After the camera and the external monitor are seeing the picture the same - then adjust the camera LCD to approximate the image on the external monitor, and all should be set to pretty much the same standard - your eye.
The key is setting something as the standard and adjusting the others to that. Changing values on all three is just chasing your tail.
F23s often have an engineer wired to them to set up all the peripherals, AF 100s are low budget so you takes your chances. Perfection is not easy, and may be impossible. TeeVee used to have a standard, as did motion picture projection. Now folks screen product on phones, computers, HD, CRT, LED, LCD, OLED, digital motion picture screens and who knows what else.
Don't even get me started on the battle of 7.5 IRE setup versus true black ( as if such a thing existed outside a vacuum).
Oy, a bugaboo would be easy compared to this stuff.......

[Josh Bass]

I think i miscommunicated something somewhere here.


I'm not concerned about the cam's lcd, just like on any regular camcorder. I trust them only in the roughest sense, otherwise, we'd never need zebras or external monitors!

I was only trying to get a production/field monitor to show an accurate representation of what a DSLR is seeing, like any of us would with a camcorder (EX1, HVX, XL2, Varicam, etc.).

With a regular camcorder, you simply run the cam's bars to a monitor, set up the monitor correctly to those bars, and BAM, your monitor SHOULD reflect what the camera is doing in the truest sense.

As for matching reality, that's all very subjective. What if you want a warm look? A cool look? etc. Cam menus and such can be tweaked to those purposes, or white balancing "incorrectly," etc.

All I'm concerned with in this thread is simply getting a monitor to be accurate to a video DSLR. Looks like our first day's results came out okay with the monitor calibrated to the XL2.

Anyway, thanks.

[Denis OKeefe]

With a regular camcorder, you simply run the cam's bars to a monitor, set up the monitor correctly to those bars, and BAM, your monitor SHOULD reflect what the camera is doing in the truest sense.

BAM - the monitor reflects what the generated bars are doing - the camera is not even in the mix at this point.

That signal bypasses all the "camera settings" so even if the generated bars look great on the monitor that is no guarantee that the camera image will look great too.
This is at least a two step process, set the monitor using generated bars, then adjust the camera settings by comparing a camera generated image on the monitor to what you see in reality. If you want then try to match that camera lcd as close as possible, but as you say trust it in the roughest sense.
The good news is that the camera settings don't seem to drift, once you have it set as you like you shouldn't have to mess with it.
If you have access to a waveform monitor and vectorscope you can be more accurate and consistent.
Here are two links to downloadable charts that apparently can be loaded on a cf card and may be useful.
HD Testchart
Monitor calibration and gamma

[Jon Fairhurst]

I wish the 5D2 had calibrated bars. Over HDMI, there is a high and low contrast version. Over analog, the offset and gain of the output aren't guaranteed. By generating the bars from the camera, you can include these factors in the calibration.

Some monitors have false colors. By calibrating the monitor to the camera, the false colors now have meaning in relation to the range of the sensor. The false colors then allow us to set exposure correctly. If uncalibrated, exposure based on the monitor's false colors could be incorrect. We might end up with clipping or an under-utilized dynamic range.

I've tried to modify a camera-generated photo to show colorbars, but the camera won't display the image. Does anybody have a RAW or JPG image of accurate colorbars that will play back from the 5D2?

[Josh Bass]

Thanks, I will look at those.

Regarding the next post, there is a jpeg image referenced in another forum that those folks recommend for just such a purpose.

However, I tried loading onto a card and calibrating monitor to it. Monitor brightness ends up way too low. . .image looks way darker than what I believe cam is actually seeing especially compared to cam lcd (if you put any stock in that) and my own plasma. I don't trust it.

Also took that image into FCP to look it on scopes, though I get a little hazy on whether this is valid, so forgive me if I'm wrong. Anyway, on FCP scopes this particular jpeg is off by about 3% on all black/white values. . .7.5 pluge bar reads 10%, superblack (3.5%) reads 5%, etc.