ISO/ASA Ratings for the XL1s

I'm used to lighting for f-stops. if I set my shutter speed to 1/60, I'd like to know what the consensus equivalent ISO rating for the new XL1s is so I can get my f-stop for lighting film style. ANyone happen to know what it is. I can't seem to find it anywhere.

PS- I read that the XL1s has about a 2 stop difference from the XL1

-J

[Andre De Clercq]

The answer is not straightforward...Gain settings on yr camera, S/N ratio, accepted film grain.. all are involved. If yr camera has a serious light sensitivity spec (not those "low light" lies) say something like "f11 @ 2000 lux" you can calculate the ASA/ISO value you have with standard camera settings (no gain tweaks and set at full dutycycle=1/60 for NTSC)) and excluding noise/grain considerations. As far as I remember f11/2000 lux must be between 400 and 800 ASA equivalent. You better take a white piece of paper put it on yr lightmeter (photocamera) and see what the ambient reads at 1/60. Take the same piece of paper, put it on yr camcorder in the same ambient light and see what the results are for 1/60 and (manual) full output (zebra level).

[Jeff Donald]

ISO/ASA ratings for CCD chips aren't that simple. There are several articles over on the XL1 Watchdog site http://www.dvinfo.net/canon/articles.htm go down the page to camera head and the second and third articles are about ISO/ASA film rating. Film behaves in a stable manner throughout the normal exposure range. However, during long timed exposure the rating changes. Double or even triple the time needed for proper exposure. This is known a reciprocity failure.

A similar phenomenon occurs with CCDs, but much earlier. A simple way to put it is, under exposure does not happen in a prefectly linear fashion. Shadows tend to block up much earlier. Light meters for video are best used for determining lighting ratios on the set.

Jeff

[Andre De Clercq]

Very correct Jeff, but very long exposures (several seconds) do not apply for normal video work. Only in astronomy (CCD) captures, those non proportionalities apply.

[Jeff Donald]

It's a similar phenomenon that occurs during normal exposures. The CCD does not perform in a linear fashion when subjected to under exposure. Normal photographic light meters assume a linear response curve based on film. CCD respond differently.

Jeff

[Andre De Clercq]

Jeff, are we talking about under exposure (yr last post) or long exposure (yr first post)? These are two different situations. CCD (photodiodes) are basically linear devices (1 photon=1electron/QE). Film needs 3 to 20 photons before a grain gets a latent image, so is nonlinear at the very low photon impact. Both have of course limits (saturation..) If underexposure is the game it's the above non linearity which counts. Long exposures are infected in CCD by the "dark current" and some device leakages. Neither the underexposure, nor the long exposure come into play for video. The limited dynamic range of video images and the viewing conditions (display brightness settings, ambient light..) do not allow strong underexposed scenes on one side and the integration time story is non existing because (for other reasons) very long exposure is not available in video
There is an easy way to contradict this myth for videowork. Take whatever CCD cam having manual controls, set the longest possible exposure time togheter with the highest F-number. Point the cam to a scene which gives 100% video for this condition. Step down the exposure times coupled with the coresponding upstepping of the the f-number. The videolevels will remain perfectly constant throughout the whole exposure/f-number range.

[Jeff Donald]

If all you want to photograph is 18% reflectance a light meter will work just fine. The problems starts when under and over exposure are induced. The response of the CCD is not as linear as film. This may be in part be due to the 8 bit recording method DV imposes.

Take a Kodak grey scale and photograph it with your camcorder. Then under expose and over expose the same grey scale. The grey scale is logarithmic, so any change in lighting should be logarithmic (1 photon=1 electron). A change in brightness by one stop is a factor of two. If I increase the exposure the the scale should get brighter in a logarithmic manner. A straight line.

However, the response is not in a straight line. If you increase exposure the whites clip in a non linear fashion (limits of the 8 bit format?). The gray scale does not perform like film. The darks (shadows) do a little better. The darks will show greater latitude than the whites, but still non linear. These whites and darks are the equivalent to D-min and D-max in film.

Try it. I didn't believe it at first either. I did these test back in August after Bill Ravens (member) and I discussed this very same topic. A conventional light meter doesn't work.

Jeff

[Andre De Clercq]

First under exposure...then long exposure...now.. over exposure..What you experience is the much higher exposure lattitude of film vs standard CCD cams. (there exist others having 9 f-stops vs 6 to 7 for film!). This does not relate to the 8 bit DV sampling but to the ccd properties. Standard camera's are not made for "artistic" over and under exposure. Hence lightmeters can only be used for standard scene exposure and are predictable in these cases. If you could get the Technical papers from the SMPTE Jounal, Januari 1996 titled "The all digital camcorder- The arrival of Electronic Cinematography" it would be very helpfull in such discussions. For another believer in the ISO story see http://jkor.com/peter/700lightable.html

[Jeff Donald]

The world is not a perfect 18% reflectance. In your example there is no under exposure or over exposure. No artistic expression. Just straight linear exposure. If you want a linear exposure use the built in meter (zebra pattern). Even in your linear example the grey scale does not perform in a linear fashion. It is not a mater of film vs. video and contrast range. When the whites get near 100% the values are not measuring in a linear fashion. The same can be said for the blacks (dark grays) but it is not as sharp as the whites.

You are absolutely correct that standard cameras do not have the features and adjustability of high end cameras. It is the use of these high end features that enable digital cinematography. Features like, adjustable gamma, 14bit signal processing, knee correction, black and white shading compensation and variable linear matrix to name a few. The question is about the ISO/ASA rating for the Canon XL1 (original question). My grey scale exposure experiment was performed with an XL1. Within the video contrast range of black with no detail (D-max) to white with no detail (D-min) the scale is non-linear. I do not find a photographic hand held light meter practical or accurate for determining exposure with the XL1. I prefer a waveform monitor, zebra pattern and calibrated broadcast monitor for determining correct exposure with an XL1.

Jeff

[Adrian Douglas]

There seems to be a general ideas it's around 160asa

[Andre De Clercq]

I suppose, when we talk about lightmeters/measuremant, we are talking about incident light measurement. I would appreciate if you would explain the mysterious differences between the two "lightmeters". Both are photodiode based, non image forming (integrated lightflux measurement...). I am interested to hear your story why and how, after ASA calibration (my first post) they two would output different settings as a function of incident lightflux (Lux), if both are used within their dynamic range limits (which is the purpose for consumer video shooting).

[Charles Papert]

Don't know why I would jump into the middle of this one, but...

The only light meter I see useful at all for video shooting is one that reads in footcandles, which allows the gaffer the opportunity to light a set to pre-determined level rather than a specified f-or t-stop. It also allows for other ratios to be set, albeit from a mathematical standpoint which doesn't allow for the particular gamma curve of the given camera (e.g., a given amount of fill will create more or less contrast depending on which model/format of video camera is shooting it, due to the differing chips and processing). I agree with Jeff that a good monitor & waveform are much more powerful tools than a light meter for digital shooting, in that the light meter is reduced to a rough reference rather than a virtually absolute one as when used for film exposure.

As far as using a 18% grey card, my personal recommendation is to use a spot meter read from alongside the camera rather than incident meter placed on or next to the card, since the angle of the card to the light can significantly change the indicated vs actual exposure. If an incident meter must be used, use the flat disk and try to keep the card as true to the plane of the camera as possible (not angled towards a light source off to the side).

[Andre De Clercq]

Agree about the "rough reference " from lightmeters.One has to know, how to interprete the results depending on the scene's contrast ratio's vs the limited lattitude /dynamic range and gamma of comsumer camera's. Also agree that waveform monitors and a calibrated displays are the better solution. If one wants to go even further in the ideal exposure game, "exposure histogram" analisys/optimisation is the way to go...

[Jeff Donald]

You have two basic types of light meters. Incident meters measure the light falling on the subject. Reflectance meters measure the light reflected by the subject. Reflectance meters are calibrated to an average scene, usually 18%. However, reflectance can vary, as Charles points out. Even the slight tilting of the subject (or gray card) can affect the exposure reading. Incidence meters do not rely upon reflectance. They measure the light before it hits the subject. The difficulty with incidence meters is they have to be in the exact quality and quantity of light as the subject. They work great in a studio, but are problematic in the field. Not convenient for getting a reading on the other side of the Grand Canyon.

Most film cameras have a reflectance meter built in. The meter is used by the camera to adjust aperture and shutter speeds when in various auto modes. I doubt the XL1 even has a meter. The various auto exposure modes are set by the video level and an algorithm used by the CPU (my best guess).

In my experience incident meters work best in fairly even lighting (lower contrast). If the incident meter is held in the light falling on a scene and the scene contains harsh shadows the meter will probably not give a correct exposure. In this situation the reflectance meter will measure the entire scene and average the readings from both light and dark areas for a better average.

I also find that some scenes do not meter well. A white sandy beach ( after all I do live in Florida) is difficult for both types. The reflectance meter usually under exposes the sand. My incidence meter gets it too light, no detail in the sand.

Jeff

[Andre De Clercq]

The "reflectance meters" in videocams (and digital photocams) are indeed based on the image properties (video AGC). Simple algorithms just average the video signal and are as such behaving like a standard reflection meter, interpreting the average reflected licht towards the camera. In AE mode, those simple integretors are very sensitive to the image's histogram properties and often result in (soft) clipped image when there is a dark background against a well lit (relatively small) foreground object. The well known backlight compensation, compensates for the inverse situation. More sofisticated systems are (weighted) multizone based and/or semi peak detection based. Video AGC concepts, quite often used in VCR's and TV's (projectors) in order to produce "bright" images, struggle with the same tradeoffs. Inventive marketing people pretend that their systems are even fuzzy logic or neural computing based.