5D replacement announcement next week
- Thread starter squig
- Start date
Which is why so many people got bent out of shape when they released the C300 at $16K. Not for what it is, but for what it isn't, and what that means for the rest of the line. Basically, blah.
This is what you can expect for the 5D and video:
Full HD. Improved ISO. Better control of moire, aliasing and skew. Not eliminated, but better. 8 bit 4.2.0, probably clean HDMI 8bit 4.2.2.
And that's about it.
Last edited:
squig
Veteran
woohoo built in iTunes!
squig
Veteran
They're probably saving it for the 4k DSLR
Jade Lane
Active member
On swappable CF Card Playlists
squig
Veteran
Easily worth the $1000 upgrade. I'm going to start writing a moire anthem.
cowpunk52
Veteran
This 'theory' is getting bandied about quite a bit, but is actually incorrect. Barry Goyette, a C300 owner and the shooter behind the infamous "By the Light of the Moon" video shot by moonlight at ISO 16,000 and 20,000, had this to say on another forum regarding that claim:
"Yeah. No. those reports were completely incorrect in my opinion. It came from an early test video where they had a 1dmark4 (not the 5d, and a somewhat faster sensor I believe) at 6400 and compared it to Log-C at 20000 on the C300. The problem with the test is that the poster wasn't really thinking about what he was looking at when he made the claim.
The 1dmark4 gamma and the Canon C-log couldn't be more different. the 1dmark4 pushes the mid-tones up at the expense of the highlights and increased noise, Log-C protects the highlights by pushing the mid-tones down keeping noise low overall. As the camera seems to have extensive range in the shadows, there isn't much penalty with this approach...especially up to about 6400 on the c300. At 20,000 it gets a little dicey, but, well...it should. So when you look at that test footage, taken in a normally lit room with the 1dmark4 at 6400 and Log-c at 20,000...they look kinda the same, because the gamma curves are so different, but in terms of dynamic range they aren't even close. Try to adjust that 1dmark4 footage and you'll know what I mean. Alan Roberts in his review stated that the c300 iso rating is correct.
For my tests I was using the normal 1 setting which is closer to the canon dslr gamma but still darker, and also in C-log.... there was a slight benefit in terms of noise to shooting with the Normal 1 setting versus trying to stretch the c-log up in post. As the Light level was so low, it didn't really makes sense to use Log-C as nothing in the scene is getting even close to testing the highlights, except for the passing cars --(which we hadn't even planned on, and there were few of) If the light level was slightly higher, then I'd probably stick to c-log as it would have lower noise and better DR.
Anyway...what I can tell you is that midway through our shoot I took the lens off the c300 and put it on the 5d. Turned the iso up to maximum, and couldn't see a damn thing (which, as Jon mentioned, matches my previous experience.)"
This is strange, but you know what? Having messed with the C300 my conclusion was just this. C-Log/Log-C was kind of difficult to work with and understand at first blush, and wasn't doing what I expected it to do as a "log" profile. Indeed, it makes sense that the mids were shoved way down, and it makes sense now as to why the footage from the night segment (3200 ISO) was a better grade than the daylit stuff (850iso).
So basically, you want to shove the 850ISO exposure down pretty far in daylight, since the sensor can reach into shadows a lot better than it can into highs?
Also, I think it was probably the best choice for Canon to make as far as how their log behaved, because otherwise you're forced to cut light before the sensor.
Let me know if I'm off, CP, but that's what I concluded, anyway.
So basically, you want to shove the 850ISO exposure down pretty far in daylight, since the sensor can reach into shadows a lot better than it can into highs?
Also, I think it was probably the best choice for Canon to make as far as how their log behaved, because otherwise you're forced to cut light before the sensor.
Let me know if I'm off, CP, but that's what I concluded, anyway.
cowpunk52
Veteran
Definitely on the right path. At any ISO, the C300 with C-Log will see further into shadows than it will into highlights - but by how much varies according to what your ISO is set to.
If you shoot below 850ISO, then you'll definitely want to shove your exposure way down. But at 850 and above, all the way to ISO 20,000, the DR of the C300 will reach at least 5.3 stops into the highlights and 6.7 stops into the shadow - which is actually a very nice balance. Below base sensitivity (under 850), the camera starts very gradually seeing further into shadow and less into highlight. This peaks at the lowest sensitivity of ISO 320 with 3.9 stops over and a whopping 8.1 stops under.
This is all, of course, according to Canon's own released specs on the camera. They have not published the DR specs of their normal gamma modes, which will behave far differently from the C-Log. Non-log gamma modes on any camera boost midtones artificially. If you expose caucasian skin for 60-70 IRE in non-log gamma, then you'll probably want to expose it around 40-50 IRE in a log gamma. This all goes out the window, of course, when you start shooting below base sensitivity.
[
Last edited:
Thanks a lot for that. That's pretty sick about the 320 being able to grab lows like that I did not do that in my shooting, because I assumed that you were pretty set with highlights, however after viewing the footage I realized that was not the right way to approach this camera. That also means shooting C-Log is pretty much like committing to a post workflow where a color session is mandatory. 850 ISO being the base (clearly show in the menus, smart Canon) is definitely accurate.
In this case, the normal gammas (or setting up one that plays well) is going to be something of great value. KNowing the tool beyond that, say the 320 thing, it would've been a tad smarter to underexpose just slightly and shoot 320 instead... being that we were in the 'woods' and the contrast scenario was a bit of an issue.
There's a lot to learn about this camera, I want to play with it more.
It also goes hand in hand with the idea that if we had film mode customs with the GH2, we may be able to use a linear curve that shoves the upper half of the curve down so we're not forced to use crazy amounts of ND or iris stops to solve the limited DR issue. Same with the Canons, to be honest.
Yep. That's my fear. I'm afraid that the long awaited 5d Mark III will not bring much to the table. If the 5d Mark III IQ = C300 then I will be THRILLED. So do I think Canon will do that....no, I don't think they would do that...and certainly not "on purpose." I'll cross my fingers just in case. So that means it will be slightly better than the 5d Mark II....slightly worse than the C300....there simply isn't any where else for it to go. Bummer....but we'll see.
Cheers!Fair enough. Here's to hoping that the ISO is TRULY 20,000....because it looks AMAZING for that level. All of that said, although it may be debated that the DSLR mappings are "off" or not.....one thing is for certain: The 6K FS100 spanks it rather handily in low light head to head, no matter what you call the ISOs and/or DBs.
Certainly we'll just hope for improved low light from the 5d Mark III whatever the case with the C300. Thanks for the perspective! Cheers!
Rick Burnett
Section Moderator
So this graph has never made sense to me, the way that it's labeled it says that in Canon Log mode, the camera basically sees a wider dynamic range that it puts into output. Why would they limit Normal 1 to a lower input range? That makes no sense. Instead, the curve would just be different squashing the top and bottom to leave the dynamic range for the midtones.
Rick Burnett
Section Moderator
I've often wondered this question, if cameras have different response curves (input->output) how do we measure digital ISO? We had this same problem when comparing the 7D/AF100 a long time ago. If I exposed for the hilights, the mid tones were off, in both directions. Opened my eyes to how complex this stuff gets when trying to compare since there are SO many little differences that add up.
Well, when I say accurate, I'm saying "this meter told me exactly what I thought it should be" But you do raise a pretty good point.
Ted Ramasola
Veteran
I'm afraid that the long awaited 5d Mark III will not bring much to the table. If the 5d Mark III IQ = C300 then I will be THRILLED. So do I think Canon will do that....no, I don't think they would do that...and certainly not "on purpose." I'll cross my fingers just in case. So that means it will be slightly better than the 5d Mark II....slightly worse than the C300....there simply isn't any where else for it to go. Bummer....but we'll see.
I'm also thinking about this; that there are a considerable number of 5D mkII shooters already using Magic Lantern, Canon also has to consider that. That whatever they improve on the Mark III has to be better than 5D mkII + Magic Lantern Unified.
5D mkII+ML Unified already has these and more;
-Headphone monitoring
-Focus peaking and other focus assist features like;
-Magnified window
-False color
-Histogram while recording
-Waveform while recording
-Clear screen HDMI out for use with ext recording device (Im careful not to use the term Good HDMI out signal because I'm not sure how good it is)
-Split audio channels with independent controls allowing use of internal and external mic simultaneously
-HDR video (albeit tricky to use with fast motion, ok with 60D, 600D due to their 720 60P)
-customized crop marks
-Built in Intervalometer for video
-Variable frame rate from 3 to 31 fps which is good for clean lowlight/night shots or for motion blur effects
-Programmable rack focus
Also, for much less cost than a new DSLR, the mosaic filter has solved Moire and Aliasing on the 5D mkII
There are also several features for stills like focus stacking and bulb ramping.
So with these features in our current 5Ds I for one am looking for significantly more improvements in the new 5D mkIII than just merely improving on a stock mkII.
Last edited:
squig
Veteran
Sony/Nikon sensors are better than Canon's in low light and I don't expect that to change too much with these 2nd gen HDSLRs.....They might not even be good enough to be worthy of the 2nd gen tag next to a GH2 image wise (I hope not). High ISO performance should be good but it won't be as good as the D4, probably barely as good as the D800. Canon have to do something about the chroma noise. The Sonys/Nikons don't have any.
The FS100 has a couple of issues that have put me off it (sensor reflections and highlight rolloff) but I might get one just for 60p 1080p if the MKIII doesn't have it.
Thanks Kholi! Good to know.
Yep. Magic Lantern is DA BOMB! Will the 5d Mark III be able to be hacked? Canon had better hope so because they've added some great features....
I'm sure you're hoping that Canon surprises us too?
Cheers Buddy!Indeed. Sony/Nikon (same sensors btw)....yepper, better in low light. Will Canon surprise us? Prob not. Let's hope so, but I'm not expecting it. FS100....indeed, not the greatest image, 1080/60P is sweet though (like you said), and that low light performance is nice if you need it. Cheers! Let's all hope to be surprised, but not cry if they don't....agreed? LOL!