Compared: F3, Alexa, Red, 35mm, 16mm and 1D.

Ok I hate to have a negative comment on here but... Its not really showing the F3 in its best light. it seemed a lot of the footage from the F3 was improperly exposed. I have shot with the F3 in super harsh sunlight conditions and had much better results. Its not really showing the full DR of the F3. In your test the highlights are all blown, but I know the camera can handle them.
 
Last edited:
When I saw the ungraded images I thought a couple were over exposed and a couple under exposed. My thought is that whenever multiple cameras are used the operator can't cater to each one.

My general impression is that the F3 is so light sensitive that it should probably be under exposed a half stop to a stop from what you think it should be a first blush.
 
Agree with Giuseppe. It holds up to any digital camera out there next to Alexa and can look very film like.
 
I have the same gut feeling as Giuseppe that something was wrong with how the F3 was shot, and I know how hard it is to when shooting tests especially with 6 cameras.

I was wondering what the set-up of the camera was and if perhaps you shot on a cine gamma and slightly overexposed perhaps using 100% zebras as a guide which can be hazardous in cine gamma. If so that could have put a lot of your exposure into the top of the scale where the gamma is rolling off radically and there is lots of lattitude compression going on. You are much better off under exposing than overexposing an F3 for sure. Did the underexposed stuff hold up better?
How were you judging exposure?
 
Thanks for posting this. One of the best "compare" shoots I've ever seen. I wasn't surprised by any of the. For this application (posted to Vimeo), the only camera that was "out classed" was Super 16. Totally got it's tail kicked in this scenario. The DSLR footy was a "give away" when the moire showed up. No surprise there either. I agree that the F3 was over exposed, but I think that is characteristic of it's light sensitivity. I don't *blame* them for over exposing...but it was.

Interestingly enough, this actually doesn't change my view of anything that I already thought....EXCEPT...that the DSLR did *MUCH* better than I expected. I picked it out right just as easy as I picked out the Super 16, but for it's cost, the DSLR continues to perform unexpectedly well. NOT a professional tool by any measure....but an Indie dream no less. (Enough of that)

I'm so happy for Alexa....what an amazing tool. 35mm has met it's match. Crazy stuff.
I'm happy for Sony. Although I didn't see it in this video...I have no doubt that it will perform *MUCH* closer to Alexa....and ~maybe~ (in this application) just a *hair* less than Red. I could be wrong (this vid it doesn't look that way)...but I'd bet that the F3 will show it's stuff soon enough.

I guess that would be another surprise (from this video), the gap between the DSLR and the F3 wasn't nearly as wide as I think it "truly" is. Anywho...good stuff! Thanks for posting!!!
 
They did note in the Blog that the F3 was a pre-production model see below, so that may have made a difference if Sony made some tweaks to the shipped cameras....although I think the F3 looked great, from what I have seen it can perform even better.

"Sony F3. This was a late addition, but I thought it would be worth while. It was an engineering sample, so it wasn’t finished and we recorded natively to the internal recorder."
 
Looks to me like they measured exposure with a light meter on the F3 footage instead of Zebra/Histogram.
that's asking for trouble.
 
I think the major difference you are seeing with the F3 and the DSLR are the result of recording to an 8 bit format. Grading choices also did not really demonstrate the full range shadow and highlight detail differences that are possible. But that was not the intention of the test anyway.
 
Razz16mm said:
I think the major difference you are seeing with the F3 and the DSLR are the result of recording to an 8 bit format. Grading choices also did not really demonstrate the full range shadow and highlight detail differences that are possible. But that was not the intention of the test anyway.
Click to expand...

I have to completely disagree. I'm sorry but just because its 8 bit doesn't mean thats why its blown. I have tons of 8bit footage from the F3 in super harsh sunlight and it looks perfectly fine after grading. I believe they just clocked the light exposure wrong and over exposed. Specially if they weren't using any high end large calibrated HD monitor on field. This has nothing to do with 8bit grading. I've graded tons of stuff in 8bit and pushed it till it fell apart, but thats not what we're seeing here, its just over exposed.
 
I am finding the same thing with my FS100, I need to underexpose a bit to get the absolute best of it, so I can see the same thing with the F3. The sensitivity is just astonishing.
 
It would be nice to see some info as to how the F3 exposure was determined because something was obviously wrong. I also agree with Giuseppe, sure 10bit 4:2:2 will be better but you can grade 8bit 4:2:0 F3 or EX files quite well as long as your not doing major surgery on the stuff. Shoot it correctly in the first place and you will be fine. I've had much more serious problems trying to grade Canon footage. I find it very difficult to buy that a Canon will grade better than an F3.
 
And I'll repeat this here because it's worth repeating. Exposing for a certain element in a scene is NOT always getting the best exposure for that scene. The problem is, where are you on the curve? Are you shooting with similar curves? In all of the comparisons done between the AF100 and the 7D, the first thing I noticed when people were exposing for the skin is the AF100 would blow out first, however, the shadow detail on the AF100 was SUPERB. If you go back and expose to the highlights and shadows, then it was clear to me that the AF100 not only had a better DR than the 7D, once corrected in post to fix up the skin tones to where they were wanted, the overall image DEFINITELY surpassed the 7D. I have both a 7D and a AF100 so I was able to perform this test on my own.

This is the problem with these camera comparisons, someone who knows a camera intimately will instantly be on here saying, wait a minute, I *know* the camera can do better than that.

Now some people make the distinction that some cameras just work better out of the box at handing shooting better, and that is true, and valid. BUT, when these cameras have so much tunable features, for a real comparison, it takes someone a good deal of time to:

(1) Understand the differences in the cameras
(2) Tune the cameras for optimal performance for *your* shooting style *or* limitations in the system that mean you need to shoot a little different.

There isn't some golden infallible rule that says one method of determining exposure is right or wrong over another, what matters is the final results.

If I were (and sometimes I do) testing out the handling between two cameras, I'd be doing a static shot of a scene that really pushed the bounds of ANY camera BUT was realistic to what I expect to shoot (and this is not in reference to the test, just in general). And I'd probably shoot it with multiple exposures so in post, I could examine the results and determine what needs to be done to get the best image. I might find other things that I don't like, so I'd go back, tune the cameras more, and shoot again until I understood EXACTLY what it is that I was seeing and what I can do to fix it.

I did this with the AF100 with regards to some noise I didn't like in the dark grays. I spent a few days shooting over and over and changing settings till I got an image that I absolutely thought was great. It was *NOT* like that out of the box. I've been pretty consistent in saying that I feel the AF100 default settings do not do the camera justice, but that is just *my* opinion. Others do not agree with this, and of course, that is valid as well.
 
Thomas Kist said:
Looks to me like they measured exposure with a light meter on the F3 footage instead of Zebra/Histogram.
that's asking for trouble.
Click to expand...

I have noticed something strange with this too. On my AF100, using a light meter, i consistently get over exposed images when using the readings from a meter, and it's generally by half/one stop over... Strange..
 
I don't know how they did the post for the F3, but it's worth mentioning that re-wrapped quicktimes from the native XDCAM have their values mapped badly.

That is, you can bring these re-wrapped Quicktimes into FCP or even Resolve or Color, bring down your whites, and more detail will come out of nowhere. The headers in the re-wrapped QTs are applying white and black points that are clipping on both the top end and bottom end.

This is the same problem the Canon's suffer from as well using Log & Transfer in FCP. The QTs are clipped by default, but apply a filter or color correction and it can be bright back.

So to my eye, in this test we're talking about, it looks like they shot Rec709 gamma and then did not try to recover the clipped bits in the computer later.
 
also the color management for the grade was through and through for projection. not sure if it was graded for p3 out of a projector, and they created a dcp for it. but looks like that might have been the case, rather than staying in a 709 space and projecting that way. effectively that can cause some issues with what we are seeing in vimeo, and what they are actually see on their monitors and projection.

but there's no telling how much information was really in the native files either way, if it was clipped it was clipped. these tests aren't perfect, they never are and never will be, it's just a comparison with the best they have at hand. no biggy, there will be plenty more, i've seen f3 handle highlights decently on native xdcam, it does an OK job imho. not a fan of what it does to the shadows either, or the color reproduction of xdcam. through on a 10 bit recorder though and I see a major difference.

than throw in the log, and we are in a whole new world.
 
Nate Weaver said:
I don't know how they did the post for the F3, but it's worth mentioning that re-wrapped quicktimes from the native XDCAM have their values mapped badly.

That is, you can bring these re-wrapped Quicktimes into FCP or even Resolve or Color, bring down your whites, and more detail will come out of nowhere. The headers in the re-wrapped QTs are applying white and black points that are clipping on both the top end and bottom end.

This is the same problem the Canon's suffer from as well using Log & Transfer in FCP. The QTs are clipped by default, but apply a filter or color correction and it can be bright back.

So to my eye, in this test we're talking about, it looks like they shot Rec709 gamma and then did not try to recover the clipped bits in the computer later.
Click to expand...

I would agree with this and with Rick. If the file as recorded is not clipped then the information can be recovered with proper grading. Gamma issues with Quicktime and FCP are well documented. Seems like most video cameras are calibrated to expose middle grey with about 3 stops of headroom before clipping so there is less shadow noise. I usually find myself shooting video about 1 to 1-1/2 stops under to better preserve highlights and get a richer tonal range. The noise trade off is not too objectionable in most cases unless you are already shooting in very low light conditions with gain boosted in which case you don't have the latitude to play with.

Seems like no camera looks its best under the limited conditions of this kind of testing. They are fun to watch, but the DP setting up the shots probably gets a lot more useful info out of the process than we do watching a vimeo file.
 
Last edited:
I guess I just dont really see the reason for these "shootouts" if they are not a representation of what the camera can really do. I shot many tests with my F3, but I wasn't comparing it to another camera to say if it was better or not. Once you add other cameras, you are comparing which then means you have to have them all on the same plate.

I can, without a doubt, say that I could get the F3 to match that Alexa footage within a few minutes of handling that camera. Also if you are unfamiliar with a camera, how can you really do a test with other cameras? I'm just not sure whats trying to be achieved here. There was just too many inconsistencies within the images, I've seen 16mm handle much better than that as well. Stock and processing has A LOT to do with what was shown. I've seen 16mm that looks like si2k footage, and si2k footage shot right along side it and it was hard to tell the difference between the two ( I wish I saved the link).

It just looks like you guys got a bunch of cameras and shot, but there was no real tweaking, which means theres no real reason to show a comparison. No one is using these cameras like this. most likely the only one that would be close to what you'd get from a normal shoot was the Canon's considering they dont have much (or any) tweaking to their image so right out of the box, thats what you get.

But an F3 has a huge amount of color, gamma, knee, and many other options you need to dial in to get the image thats most pleasing as do most professional cameras, out of the box they dont wow anyone. Every DP knows you must dial in the best for the camera and to achieve the look. Thats what DIT's are for and why the good ones are the hardest once to get on your set.

The best camera comparison test ive seen yet was the Zakuto shootout (now I have to say I think it was completely a swung test and everyone involved was making some kind of $$ of DSLR's so I dont believe it was fair at all BUT...) They matched the footage beautifully. There was no huge difference between the cameras because the time was taking to setup properly and make corrections for exposure and properly grade in post as well. Thats the only true way to compare.

Best way to do a shoot out is with DP owner/ops who know their cameras well, and put them behind the wheel of their own camera, have them shoot only their camera. If I'm unfamiliar with a camera, I'm not qualified to shoot a comparison test with it, thats for sure. Sorry I'm just kinda tired of "tests" on the internet that dont really give any information or background. It just misleads the public as to what they are really seeing. Theres a lot of young guys and gals out there who are aspiring dp's and its important they learn right if we are to ensure the future of our artistry and craft.
 
I would say:

a) no test is ever perfect.

b) imperfect tests still tell us things.

I applaud those who take the time to do them. I've done them. They are imperfect, difficult and time consumming. Interpretation of what the test really tells you is the important part.

Of course it would be great to test them all yourself. However, I don't have access to all of them. I bet most people here don't either.

Don't get hung up because 'your' camera had an issue. So the F3 was over exposed. We can all see that. 1) We learned it's better to under expose a stop. 2) It still compared very well to the Alexa, film and the Red at a price I can afford. :)

GP, you're finding a glass half empty. I'm seeing it as half full.
 
You must log in or register to reply here.
Back
Top