LoganMackay
Well-known member
How so? I really don't get it. What kind of levels are you expecting? You most likely will not get true recorded range until the final models are shown to people like Barry.
AF100 / AF101 - initial report
- Thread starter Barry_Green
- Start date
How so? I really don't get it. What kind of levels are you expecting? You most likely will not get true recorded range until the final models are shown to people like Barry.
Homunculus
Carbonite Member
well for instance i found it troubling that in the philip bloom low light test his face was completely clipped out from a tiny match. unless i'm missing something that to me is not indicative of huge DR (which 10.5 would be, huge on my scale)
Homunculus
Carbonite Member
i recall the same test being done in zacuto and unless im recalling it incorrectly, none of the DSLR's clipped out in the way af100 did in bloom's test.
Matt Gottshalk
Resident Preditor
That was a web compressed pre-released footage.
Homunculus
Carbonite Member
what does web compression have to do with dynamic range on the highlights? I watch my own videos on web compression all the time after uploading them to youtube and there's never a drastic reduction in dynamic range enough to blow out a person's entire face in clipping from a single match.
plasmasmp
Well-known member
I found it disturbing too, but I'd imagine it has more to do with the type of gamma settings chosen in camera when he shot that clip. I'm sure with all of the new settings available on the AF100, you can change the way that the gamma rolls off in the highlights.
You are absolutely right that it has nothing to do with web compression. :beer: But we really need to see a final camera before we can dive into optimizing settings.
Paul Vicente
Well-known member
I have a lot of experience with AVCHD, and it tends to clip highlights if you are not careful.
Michael Olsen
Veteran
Homunculus
Carbonite Member
Ok some of you guys may be missing my point. Compression has nothing to do with it, every other camera in the world has been tested with the same test and never in my life did I see such poor results.
Here I made this comparison for you all
please tell me why the crappy DSLR nikon at 25,000 ISO still looks much better than af100. why does canon 1d at 10,000 ISO look worlds better since according to some the explanation is that the higher the ISO the lower the dynamic range. well look for yourself, these "mere dslr's" still have good enough dynamic range at astounding ISO's to not completely clip out, why can't AF100 at just 3200? please someone explain this to me
Here I made this comparison for you all
please tell me why the crappy DSLR nikon at 25,000 ISO still looks much better than af100. why does canon 1d at 10,000 ISO look worlds better since according to some the explanation is that the higher the ISO the lower the dynamic range. well look for yourself, these "mere dslr's" still have good enough dynamic range at astounding ISO's to not completely clip out, why can't AF100 at just 3200? please someone explain this to me
Homunculus
Carbonite Member
not to mention the nasty light rings around the match in the af100 that even the DSLR's don't show. similar to GH1 when it shoots a light bulb and has those nasty rings around it when 5dmkii doesn't
Barry_Green
Moderator
You're pixel-peeping again on an unfinished product. The 3200 ISO wasn't even supposed to be tested yet, as it doesn't show any difference in sensitivity from the 2000 ISO. It's not done yet. Wait until they have a working version that is fully implemented before pixel peeping.
Homunculus
Carbonite Member
ok but according to the logic of others if 3200 is 2000 in the unfinished chip that means the dynamic range should be EVEN HIGHER. since as someone just said dynamic range worsens as you go into higher ISO's. so why does the DR seem absolutely atrocious at 2000 (3200) and doesn't compare to the DR of a DSLR at an astounding 25,000 that's what i can't come to grips with.
Keep in mind what i'm talking about has nothing even to do with ISO really because I'm not complaining about the noise or anything, so whether it's 2000 or 3200 it makes no difference as the noise is fine, I'm talking strictly about the ability to hold highlights, why are they so badly clipped from a tiny match. I simply have never seen that before in my life even in a measly DSLR that's why I'm just so flabbergasted by it
Homunculus
Carbonite Member
oh and believe me I don't have the ability to pixel peep. I don't know enough about technical stuff like you guys do. If you think you need to be a "pixel peeper" to see how atrociously bad that highlight is clipped out then I dno't know what to tell ya LOL!
I don't need a pixel peeping microscope to see that nikon 25,000 ISO footage and AF100 highlite blowout to see how they compare
I don't need a pixel peeping microscope to see that nikon 25,000 ISO footage and AF100 highlite blowout to see how they compare
Matt Gottshalk
Resident Preditor
He overexposed the image hence the clipping.
I think it's just overexposed plain and simple. I think the point was just to demonstrate how sensitive the sensor was at that ISO. It's so incredibly sensitive, it's just blowing out at approximatly 1 foot form his face. The lens was at like f1.0. It seems like what I would pretty much expect. Stop it down to proper exposure and it would be fine. I'm not sure you should expect it to be holding highlights when it's just over exposed like that.
Do we know what the shutter speed and aperture settings were in all of those shots you posted?
I can see that the Nikon and Canon shots say F1.4 and Bloom said he was using a F1.1 lens, but I can't find any other information.
Are they holding the matches the exact same distance from their faces?
It looks to me like Bloom is holding the match closer to his face, and further to the side, while all of the others are definitely holding the matches out in front of their faces.
Additionally, the 3200 mode on the AF100 is a work-in-progress, so it seems a mistake to judge anything from the footage, other than the fact that the AF100 shot is clearly overexposed.
Homunculus
Carbonite Member
hmm not that simple I'm afraid. You see that's just one still taken from the video and when he pulls the match away further from his face it doesn't look that over exposed. It clips out only when the match gets closer. "Well there's your explanation" you might say it clips out only when it's close to his face, but in the other example videos for the Nikon and Canon 1d they move the match from further away to closer also and even when it's super close to their face the footage doesn't experience such bad clipping.
Perhaps as mark said it could be gamma settinsg etc etc I honestly don't know...I just find it very hard to believe that wrong gamma settings could be responsible for this level of differentiation. I can change every possible setting in my camera and it will never effect the dynamic range by a whopping 5-10 stops as it seems in this footage.
Don't get me wrong I'm still optimistic/hopeful that this is just some sort of issue that will be dealt with whether it's not finished sensor, or wrong settings or whatever ...because Barry Green DID test this thing at 10.5 stops, but it's just an issue that greatly troubles me because it has sprung up elsewhere as well and so is not a 'fluke' issue that only Bloom in his haste put some wrong settings. Such as in the collapsed white highlights on the cheek of the model in the fantastic crews.tv test footage shot on zeiss ultra primes
Homunculus
Carbonite Member
yes barry just said the definition of 3200 being a work in progress is that it has the same sensitivity as 2000 as of right now. so guess what that means? That means it makes the blow out even worse because the blow out we're seeing is not at 3200 but actually 2000, that means the sensitivity of 2000 is even less than thought going by your theories.
yes bloom used f1.1 opened up all the way and some of the other tests were 1.4 as someone stated. is 3 measley tenths of an f stop the difference in such blow outs?
if it's "simply over exposed" as some of you say then why is the nikon/canon footage not "simply overexposed" in the same harsh way? that's what I'm trying to get at
Homunculus
Carbonite Member
this entire argument can be settled by barry green making a proper match test with the unit he's currently bogarting set with proper gamma settings, exposure etc to put an end to all the varied theories espoused here
rogar-Love(DBG):
rogar-Love(DBG):