Printable View
this may be a completly silly question but it has been on my mind for some time.
after looking at a ton of Arthur Morris' photographs in the last year or so i've noticed that many are taken at ISO 250 while i haven't seen too many at ISO 200.
i was just wondering the rationale behind the slight bump to ISO from base setting? i have a few ideas, but would love to hear the real answer if there is one.
maybe there is no real answer at all, just coincidence?
I can't speak for Artie, but I also can't say that I've noticed any trend in his ISO selection. I use the lowest ISO necessary to give me the shutter speed or DOF that I need for a given shot.
I'm with Doug on this one. I usually stay away from 'in between ISOs', though. In other words, I normally use 200, 400, 800 and 1600.Quote:
Originally Posted by Doug Brown
On Canon 1 series cameras, the only "benefit" of ISO 250 over 200 is the loss of 1/3 stop of highlight headroom. Intermediate ISO's are very poorly implemented on Canon cameras.
What is the reason for this, are "in betweens" not as good, or just a personal thingQuote:
Originally Posted by Axel Hildebrandt
Dan
The S/N ratio of these intermediate ISOs is worse than the next higher 'real' ISO. For more in-depth information you can also check Emil's responses in this thread: http://www.birdphotographers.net/for...ad.php?t=26030Quote:
Originally Posted by Dan Walker
Thanks for the link Axel and thanks to Emil for all the info
Dan
Quote:
Originally Posted by Axel Hildebrandt
Correction, the S/N is the same as the next lower "even" ISO (100-200-400 etc) for 1 series Canons (and the 5D). You get the same S/N by exposing at that next lowest "even" ISO with the same Av/Tv for 1 series cameras, with the added benefit of an extra 1/3 or 2/3 stop of highlight headroom. On xxD series cameras, you get the same S/N as the nearest ISO (eg ISO 200 is the same S/N as 160 and 250); 160 is ISO 200 pulled 1/3 stop, and 250 is ISO 200 pushed one stop. Correspondingly ISO 160 leaves 1/3 stop of the available raw levels unused, while 250 pushes 1/3 stop of the highlights available at ISO 200 past raw saturation. Either way, 200 is better.
Again it should be stressed that this applies only to Canon's implementation of the "intermediate" ISO's; for Nikon the story is different, the "intermediate" ISO's are properly implemented and there is no reason to avoid them.
Thanks for the correction, Emil!
Hi Scott I often use odd ISO settings but for a good reason and has nothing to do with quality.
When shooting manual I like ending up with even numbers like f 8.0 at 1/1600 If they vary by a third it is easier to do it by ISO. Find it easier to add an substract form even numbers. btw I have no clue why Artie uses the 250.
I do not believe Canon implemented ISO settings so poorly. But I have heard that rumor before. I have not seen any issue with intermediate ISO's.
Sometimes the higher ISO's such as 250 are required to get fast enough shutter speeds.
I find the auto ISO on the 50D appears to choose the lowest ISO that gives "1/focal length"
as do i. :)Quote:
Originally Posted by Doug Brown
I cannot respond on Artie's behalf, but I can say this:
I like the use 1/3rd stops over even ISO numbers (200, 400, 800, etc) because they look very similar in terms of noise to the lower full values, while gaining a bit more shutter speed.
2/3rd stops over look much more like the upper ISO. ISO640 for example looked almost identical to 800 on all the cameras I've used so far, while ISO500 was difficult to differentiate form 400.
Thats what I do as well. I often use 100, 200, 400 and rarely 800.Quote:
Originally Posted by Doug Brown
that strategy does indeed make shooting a little easier on the brain! thanks for the suggestion Alfred.Quote:
Originally Posted by Alfred Forns
On the 40D (I presume the 50D is similar but I haven't checked), the 1/3 stop ISO's are implemented in software based on the "standard ones". For instance, for ISO 250 the camera hardware uses the hardware amplification of ISO 200 and simply multiplies the resulting RAW values by 1.25. If one looks at the RAW data prior to conversion, one clearly sees that only 4/5ths of the levels are populated, a clear sign that the data has been stretched after quantization. This makes the histogram look nicer, but in reality you would do better by using the same shutter speed and aperture at ISO 200, and applying 1/3 stop exposure compensation in the RAW converter; there is no difference whether the camera hardware does the software multiplication of RAW values or the RAW converter does, but you keep an extra 1/3 stop of highlights in the RAW data by using ISO 200. So if you happened to want that extra 1/3 stop of highlight data, it's there for you in the RAW converter, whereas if the boost was done in-camera that data is lost forever.Quote:
Originally Posted by Paul Granone
On the other hand, for ISO 160 and its siblings, the camera still uses the hardware amplification of ISO 200, but now divides the digitized result by 1.25 to make the histogram look better. So in reality you are doing ETTR by 1/3 stop at ISO 200. In this case there is not much difference whether the division by 1.25 is done in the camera of in the RAW converter; however, if that is what the camera is actually doing, then my personal preference is to use ISO 200 rather than 160, and the metering will tell me explicitly that I am doing ETTR by 1/3 stop rather than lying to me that the exposure is correct, and I am reminded as to what is going on and the potential loss of highlight headroom.
This is why I turn off the intermediate ISO's on my 40D.
For completeness, here is my reasoning for the Canon 1Series and 5D (again I've not tested the 5D2, it may be like the 5D or they may have decided to shave costs and make it like the xxD series).
The read noise in photon equivalents as a function of ISO for the 1D3 looks like this (ignore the curves, they are a fit to a mathematical model of the noise; the points are actual data):
http://theory.uchicago.edu/%7Eejm/pi...s_iso-elec.png
There is essentially no difference between the read noise in groups of three -- a "standard" ISO like 200 has essentially the same noise as ISO 250 and 320. There is a marginal improvement of 125 and 160 over 100, but virtually undetectable in images. Read noise for ISO 500 and 640 is the same as 400, etc.
What does this mean for practical photography? Suppose the exposure (Tv and Av) was correct for ISO 640. That means the sensor is collecting a given amount of photons. If instead one uses ISO 400, you still have the same photons, and the same electronic read noise; but again you gain 2/3 stop in highlights if you should want them in post-processing; if you used ISO 640 there would be no benefit in S/N, and that extra 2/3 stop of highlights you might have wanted are gone forever.
So again, I set the custom function that turns off the intermediate ISO's on my 1D3, and if I need a bit more shutter speed I simply "underexpose" at a standard ISO, and compensate during RAW conversion.
definitely one of my suspected reasons.Quote:
Originally Posted by Bogdan Coltea
very interesting post! this is information i've never come across before.Quote:
Originally Posted by Emil Martinec
i'm curious if the same would apply to all digital cameras or would this be specific only to Canon technology?
Quote:
Originally Posted by Scott Grant
Specific to Canon and the way they implement the intermediate ISO's. In particular, what little data of this sort I've seen for Nikon (complete results for the D300, a handful of ISO values for the D3) indicates that there is a S/N advantage in using "proper" exposure at intermediate ISO's, over underexposing at the next lowest "standard" ISO.
thanks for the info Emil! i will certainly be keeping that in mind.
