a55 EVF questions

[Dr. Harout]

Barry, with your standpoint, you should opt for a rangefinder. After all the latter's VF is much purer than the SLR's...

[bfitzgerald]

I quite like range finders myself but they are more a specialist thing rather than a general purpose pursuit.

I see even Mr Reichmann is now predicting the death of OVF's too.
http://www.luminous-landscape.com/revie ... ited.shtml

Glad his film predictions didn't come true choice is a good thing. Call me old hat but I'll resist EVF's for a long as possible!

[Neonsquare]

First: Walt, I want to apologize for telling you that I start to see you as a "negative, irrational old man" - it was not meant as an insult by me, but I understand that it actually is insulting. Particularly I didn't mean to say you ARE this, but that I get this impression when reading some of your postings. I actually believe that you can be a very positive and rational and fresh-minded person - it just seems that the topics we face do not give me much chances to experience that.

And, despite the usability of the EVF in everyday conditions, there are certain things it can't reveal to the user like nuances of rapidly changing expression. Walt deals with small amphibians as a subject. I think certain things including eye blinking, vocal pouch expansion, tongue catching insects etc would be quite impossible to time with the EVF - they are not exactly easy with the OVF.

Of course there are conditions where an OVF is a much better tool than an EVF (and vice versa). The reaction time of a human is somewhere around 100ms though and I think, that we should really not speak about optical vs. digital delays as the problem: The EVF has a refresh frequency of 1/60s which is actually several times quicker than human reaction time. A difference to the OVF is the time resolution though: An optical non-discrete system naturally shows a much higher time resolution which can be helpful when recognizing harmonic movements. Tests have shown that trained humans can recognize a light flash of 1/300s so a flash of much less may not be even recognizable by human wetware. I'm willing to believe, that most time resolution problems will be much less critical when EVFs start to show 150-200Hz.

Anyway, I have got myself a nice old wooden metronome and I have been doing some tests of my personal timing - and this is of course a regular repeated movement. OVF wins of course.

Yes, but there is no reason for me to believe, that increasing the FPS from the modest 60 to 150-200Hz will make even this problems vanish.

This does not change my liking for the EVF as a working tool for extreme tele, low light and similar situations. But I do not think Walt is being irrational or negative. I do think he would find the A55 useful alongside other equipment and would be impressed by its audio quality too, despite known limitations.

Well - I tend to differ in your view of Walts attitude towards EVF or the SLT concept. You recognize a set of negative points and a set of positive points - denying or ridiculing the positive points is not a rational or positive position to any topic.

[agorabasta]

Yes, but there is no reason for me to believe, that increasing the FPS from the modest 60 to 150-200Hz will make even this problems vanish.

We'd need a synchronous process of sensor readout and EVF display at about 150Hz. Very possible with current tech. But not yet there in the a55/33.

[Greg Beetham]

I agree with Walt, 60Hz isn't a fast enough refresh rate, it's probably ok up to, maybe a 50mm lens, (75mm on APS-C) and that's only 2X roughly in perspective. You measure light in billionths of a second so there is no delay to a human eye with light transferring the image from the lens too the eye optically through an OVF, so you get a realtime transfer; if you were using a telephoto lens at 5X perspective or more, or trying to use very high magnification of 1:1 and up the last thing you need is any delay due not only to the low refresh rate but the processing time lag in an EVF. There is no way you can convince me that the processing lagtime is anywhere near equivalent to the speed of light, or is ever likely to be.
Greg

[Neonsquare]

Greg,
Real time transfer of light is not really needed, because the human wetware quickly starts to be the bottle neck. You cannot agree with Walt about the 60Hz, because he never brought up this point. He just argued that the intrinsic delays of an EVF system will always be to late for the human shooter. Being to late is not the problem, even with current implementations. Compare this with real human reaction times of 1/10s (10 Hz!) and a light flash recognition limit of 1/300s. As I said - 60 Hz is a modest value - good enough for many situations but clearly a point where development has to happen. But the reason for that is following harmonic movements. So if you actually mean 60Hz is not enough for all purposes you may mean to agree with me instead of Walt .

ciao,
Jochen

[Greg Beetham]

Yeah well ok, maybe Walt didn't say 60Hz refresh, but the thing that concerns me is; using telephoto lenses, (besides macro at high magnification), if you use a tele that's say 10X perspective that lens is magnifying the object ten times, and it is also magnifying the speed of movement 10 times in the viewfinder, also everything else is magnified 10 times, the background as you pan, your own movements are also magnified as well by the same amount. That was what my concern was basically, is a viewfinder that can maybe 'keep up' at 2X going to be good enough at 10X? I gots my doubts....
Greg

[David Kilpatrick]

Greg, this is where in-lens helps. Using my 70-300mm G SSM on the A55, the view at 300mm is becoming a little difficult to live with - shake does cause a movement which looks unlike anything seen through an OVF, where the eye compensates. Using the 70-300mm Sigma OS the experience is completely different, with a stable finder view.

The situation is reversed when shooting video. The Sigma OS does not respond well to random panning or subject following movements at 300mm. The SSS on the other hand, used with the 70-300mm, is stable at 300mm even when tracking a subject.

Solution is that Sony must implement SSS during viewing, not just for shooting. The KM A2 had this option. It might be an excessive drain on batteries though.

David

[Greg Beetham]

Now I'm wondering if SSS could be operational for the EVF either, at high magnification. The reaction time for SSS at 10X has to be very fast, depending on focus distance, angular displacement compensation etc. It may turn out that the SSS mechanism is capable of reacting much faster than the refresh rate of the EVF. ILS might be the only answer in that case, unless there is a much faster EVF in the works that is.
Who has had the opportunity to try high mag BIF with the A55? or little tweety birds jumping around in branches and trying to follow that action at high mag; I know the A55 has very good AF, but what about what you can see in the VF and snapshooting and composing using that information, and is the subject in the photo where you thought it was in the EVF?
These are things I would like to know about from users myself.
Greg

[agorabasta]

Now I'm wondering if SSS could be operational for the EVF either, at high magnification.

Greg,

You may stop wondering already The thing works, just as evidenced with a55/33/580/560 video with SSS enabled...

On the 60 FPS being somehow not enough for the EVF, I can tell you that it would be quite enough if that were the only limiting factor, that's mere 17ms.
In reality there's a delay combined of exposure, readout and processing before the image is ready for display.

Assuming that no more than one frame is being processed while another is being exposed/read, we arrive at about 200fps sensor speed for the delay to stay between 5-10ms before the data is fed into EVF display. Then it's better to drive the EVF at the same exact speed so that the data is displayed right as it's ready for display.
So something around 150-200fps is about optimal.

And btw, it's nothing impossible at all. The Casio EX-F1 from 2008 could make 300fps video at 512x384 res; that was an 1/1.7in Sony CMOS sensor.

[Greg Beetham]

I was using the 60fps=1/60sec as a mental yardstick, so if you are faced with keeping up with the shimmies from 300 or 400mm lens handheld, (400mm on APS-C = 600mm) I was expecting a refresh somewhere around that, but with the SSS as an unknown quantity working with an EVF. I already remarked before on the processing as a possible contributer to the 60fps limit, obviously that also involves downsizing from the main sensor. Looks like it's another graphics processing unit apart from the still taking one, maybe it's the same one used for the video side of things.
So what you're saying agorabasta is that SSS makes the 60fps EVF work with telephoto regardless? I got the impression David wasn't overly impressed at long fl's.
Greg

[agorabasta]

Greg, the SSS is not enabled for the EVF display when you take stills - that would drain too much juice from an already constrained system.

But it may be (and is normally) enabled when you shoot video.

And having an option to enable it for stills upon half-press would be absolutely great. (Though I feel no such need myself.)

[bakubo]

Greg, the SSS is not enabled for the EVF display when you take stills - that would drain too much juice from an already constrained system.

I wasn't aware of this. I thought that with the EVF getting its image from the main sensor that has SSS that the view was stabilized. Thanks for the info. I thought that some time back somewhere I read posts about how this was one of the advantages of having the EVF, but I guess that was wrong info.

[agorabasta]

Henry,

There's one more thing nobody seems to be quite aware of. The noise diff between a55 and a580 may be much larger than 0.5 stop caused by the mirror. If a580 is used with the OVF, its main sensor stays cold; the a55 sensor heats up. Some 30C diff in sensor temp can easily bring you another full stop of noise.

With the older a700 the additional 30C normally brings ISO800 noise above ISO3200 noise of lower temp. I'm not sure that a55 is as much dependent on the temp, need more testing in hot weather. But of the Nex5 I'm pretty sure that it's much worse than a700 in that regard - a hot summer night makes its ISO1600 very noisy pretty fast.
Hot sensors are a problem to solve too...

[Greg Beetham]

So what's the state of play then? You can have SSS 'on' for video and a stabilized EVF with a standard screw drive long lens, is that right? but you can't have that when taking stills? and you can have a stabilized EVF when taking stills with a long lens if you are using a lens that has built in stabilization...SSS is turned off presumably...have I got that right?
Greg