hyalinejim

OK, here's some happier news. I had noticed previously that when shooting outdoors in C-Log or WideDR I got the nicest colour when using the cloudy WB preset rather than daylight, shade or a custom balance. Today, I made a test indoors in mixed tungsten and daylight. First, I generated a custom white balance with a Lastolite thingy:

Then, for the hell of it, I switched over to cloudy and got this:

Obviously too warm, right? So I thought it would be interesting to see if I could push it in post to roughly match the custom WB. I did and got this:

Here's an A/B, custom on left, fixed cloudy on right:

I know it's subtle in the middle of a forum page but when I on/off these fullscreen the colours in the one on the right are much nicer for me. Look at the way the warmth of the tungsten blends into the coolness of daylight. In the custom white balanced version (left) that transition seems to go from greenish warmth to purplish daylight. But in the cloudy white balance which was brought back into the roughly-correct-range-but-not-exactly-the-same-just-what-looks-best (right), the skintones come out a lot better in mixed light. And I couldn't get the one on the left to match the one on the right just through white balance alone. So there's something going on with the cloudy preset that I find quite special.

FWIW, this is C-Log. I'm using GingerHDR's white balance tool, which I think is no longer available. Then adjust levels. Then my lovely curve.

On 13/11/2016 at 4:02 PM, kidzrevil said:

Last but not least I am seeing that the in camera sharpening is vital because C-LOG can turn your footage into mush even though it has fantastic DR.

On 13/11/2016 at 9:11 PM, hyalinejim said:

 

I am seeing this as well with C-Log... must do some comparison with other styles with sharpening whacked up.

OK, so today I wanted to set up a test where I shoot each picture style in turn with its default settings in mixed lighting in order to figure out how each of them differs in terms of

• Dynamic range and highlight rolloff
• Colour rendition
• Detail due to sharpening
• Likelihood of falling apart in a grade

So I whitebalanced the scene (a mix of daylight, tungsten and fluorescent) and set middle grey to read in the middle of the exposure meter:

My plan was to then go back and maybe tweak sharpening of some of the profiles to play with how much detail could be preserved before edge artifacting sets in, and I was witness to this holy shit show:

But Canon, I don't have a pebble dash wall in my living room! More info here:

More horrors! Little black dots caused by in-camera sharpening at 4K. It's most noticeable in Standard profile with sharpness turned up to max, but it's noticeable in all profiles that allow sharpening. I first spotted it in WideDR with sharpening at the default settings. The little black dots don't seem to be apparent in HD, but the blackening of edge outlines certainly is. The lesson here is: don't use in-camera sharpening. I'm going to send these files to Canon.

Standard, 100% view, no sharpening, no little black dots:

Standard, 100% view, maximum sharpening, lots and lots of little black dots:

I agree, not enough detail in 4k to do a 2x reframe for 1080p. Would match nicely with your C100 though. Or, if you have the storage space, shoot MLRAW on your 5D, and save yourself some cash.

@Dan Wake previously, in 5x greyscale crop mode the cropped area was not in the middle of the sensor. It was to the left. Soon that will be fixed.

Maybe shoot Canon an email or phonecall and ask them. The more they hear from customers about this the more likely they are to do something.

Not a steady hand at all - it's all down to the XC10! For some shots I turned on the Power IS - the 4k only IS option that helps with static handheld shots. But when you turn off and on again that setting is not remembered for some weird reason.

No, unfortunately. This mode is still not good for shooting styles that require any level of response in real time. However, I seem to recall that if you half press the shutter button (at least on the 5d3) you get a colour preview and slightly faster fps. This helps a bit but it's still tricky if you need to follow fast moving action.

It would be better for outdoors stuff or very bright interiors I'd say Jimmy as f5.6 in a church or function room will necessitate a high ISO at 180 degree shutter 60fps and you'll get ghosting (XC10), softness and noise.

And wouldn't you need shallow DOF for a wedding?

If not, then yes stabilisation is great and you can nail a lot of shots very easily and quickly. Also a nice cam to stick on a gimbal because of its deep DOF and smooth auto exposure.

Beware of the brittleness of the codec in HD, though, especially at 60fps.

21 hours ago, kidzrevil said:

New video I shot in Canon LOG. BTW I am in love with this new batch of lens filters I picked up ! The grain was done in post

 

Nice work @kidzrevil - what delicious filters are you using at the moment?

15 hours ago, BenEricson said:

Was messing around with C-LOG around the house. Has anyone ran into this artifact? Lens is zoomed all the way in with C-LOG at iso 500 I think. I haven't seen this on any other footage.

 

Yes, I've seen it before... and I'm sure I'll see it again! I've noticed it on single frames only.

6 hours ago, BenEricson said:

 

This was all shot 305mb/24p. If you have a 4k monitor, you should be able to play it at 4k.

... It's almost like the bit rate isn't high enough to render all of that detail with everything in focus, probably why shallow stuff tends to look better. 

 

Thanks for posting - great to see what others are up to! Some great tonality there. I totally hear you about detailed shots becoming mush. Must check out kidzrevil's recommendation to try other picture styles with sharpness whacked up. And please do post your lut @BenEricson

Here's my shots from a weekend break in Belfast - I can sense the mush in lots of them even when downscaled to 1080p. 

Camera: 4K C-Log, ISO500, ETTR where possible, WB Cloudy, Tiffen 1/4 Black Pro Mist.

Post: white balance tweak → levels → my simple curve posted as a lut on previous page.

EDIT: I think my lovely grain did not survive the vimeo compression too well

More good news (well, for certain shooters). A1ex is on the way to fixing the off-centre 5x zoom crop centering. If you weren't aware, in the "5x" grayscale preview crop mode (the one that allows you to bump up the resolution) the portion of the sensor that was being sampled was not exactly in the middle. And the more you'd increase resolution, the more off center it would become. This was a PITA for those who are trying to make precise geometric compositions in this mode as it was impossible to get a symmetrical planimetric framing, or for anyone who wanted to use an anamorphic in crop mode as you would start hitting the vignetting on one side.

It isn't in the nightly builds yet, but is on the way.

http://www.magiclantern.fm/forum/index.php?topic=17021.msg174866#msg174866

That's nuts about the new Final Cut. Glad you got it sorted @Lintelfilm. Presumably this happens with log from other cameras too. 

What is Look 5?

2 hours ago, UHDjohn said:

The EBU paper didn't test the DR in C-LOG

On page 11 there's a comparison of DR between Standard and C-Log:

I don't fully understand the DR test, but I interpret this graph as showing that the clipping point at 100 IRE is earlier for Standard than it is for, but that the final maxed out clipping point at 109 is roughly the same for both. So if you're shooting standard with 100% zebras and you're seeing clipping, there could be quite a bit of info still in those superwhites. It just appears that Standard has less DR when you switch back and forth between it and C-Log because the area under the zebras increases.

4 minutes ago, kidzrevil said:

They definitely do they are just mapped differently. For instance there are profiles that have superwhites like the cinema eos standard with a raised black then you have the standard profile that lowers shadows and highlights. Alll give 11-12 stops they are just distributed differently. EBU documentation proved this

I think so too, from reading the EBU paper and from my own experience of tinkering with the files.

I shot a lot of 4K C-Log yesterday at 205mbps and noticed a bit of chroma artifacting going on with FilmConvert. So I went back to the old fashioned way of grading and came up with a curve for C-Log that I really like and that doesn't seem to cause as much breaking up of the footage. Here it is if you wanna check it out:

https://drive.google.com/file/d/0B1exEpCRAfgFVzlUaVByczNDR1U/view?usp=sharing

I like this because it's quite contrasty but doesn't really crush shadows at all. Just set your white point and black point with curves or levels before this lut.

5 hours ago, kidzrevil said:

 Last but not least I am seeing that the in camera sharpening is vital because C-LOG can turn your footage into mush even though it has fantastic DR.

I am seeing this as well with C-Log... must do some comparison with other styles with sharpening whacked up.

29 minutes ago, UHDjohn said:

The XC10 profiles do not all have the same DR - hook up a monitor with some scopes...

Yes, but is your monitor showing what's in 100 to 109 IRE?

Quote

For  comparison,  the  same  test  was  done  using  Look5  and  ISO500  (10.5dB)  which  was  expected  to have  a  much  wider  range.  The  exposure  difference  was  again  80:1,  again  making  a  total  dynamic  range  of  about  2, 3 00:1. This is a surprise, since it implies that the  only advantage in using the Looks is in the distribution of  this,  apparently  fixed ,  dynamic  range.

Source: page 11 of Alan Roberts' EBU paper

https://tech.ebu.ch/docs/tech/tech3335_s17.pdf

2 hours ago, Lintelfilm said:

How do your cameras respond to using 100% zebras to expose to the right (just before clipping)?

On mine, all of the info is there sitting under 100 IRE on the waveform monitor as would be expected. Is your software somehow interpreting the footage as 16-235 levels? Can you upload a clip for us to check out?

Looks great! You've been having some nice weather in NY. I'm heading to Belfast today and will try to get some shots there.

Why do you prefer CLog and Cinema EOS Standard to the others? I'd be interested in your thoughts on this as I'm slowly working my way through the various picture styles. I agree that CLog is best for recovering highlight and shadow detail, but I found a lot of banding compared to WideDR. But it does have compatibility with a huge selection of luts. 

WideDR looks similar to CLog but as sharper roll off into highlights and shadows and the colours are slightly different.

With Cinema EOS Standard and Standard you can drop the ISO to 160 - although I suspect the white clipping point is the same as CLog and WideDR at 500 and shadow noise looks around the same. But people say ghosting is decreased. However, both these profiles have a video-ish saturation burnout in the highlights compared to CLog and WideDR. This can be fixed, a bit, in post by gently desaturating highlights.

Now, they also differ greatly from each other in colour reproduction. Standard seems to me closer to WideDR and CLog. I'm not sure yet what to make of the colours in Cinema EOS Standard.

Basically, all four of those picture styles offer the same clipping points, but they differ in how usable different parts of the dynamic range are: CLog is best for shadows and highlights but suffers from banding if you push it too far. They all give (sometimes dramatically) different colour reproduction. And ghosting is worst on CLog, followed by WideDR, and better on the standard profiles.

Will continue testing!

PS: For sharpness, the EBU paper says 3 is the max before aliasing begins to hit IIRC

@kidzrevil did some very quick tests of Cinema EOS Standard. I found the colours very different to WideDR. Then I looked at Standard. Colours are closer to WideDR. The level of highlight and shadow detail that can be pulled back from both of these at ISO 160 is very similar to what WideDR offers at ISO500.

So if it helps to decrease ghosting, I'm all for one of these at 160 rather than WideDR or CLog at 500.

I assume you know about the internal ND, and you're saying it's not enough in bright light. I agree with Tim that a vari ND is great outdoors, giving that elusive level of manual control. Here are my tips.

I like to ETTR  to avoid shadow noise. There are superwhites from 100 to 109 IRE. When shooting manually I set zebras to 100 and go around 2 clicks brighter in C Log and 5 in WideDR. If I'm shooting auto I set EC to plus 1 (but I wish there was an assignable exposure lock on/off button).

You can get a slightly lower base ISO by switching to gain and fine tune.

Update to latest firmware to minimise RS in 4K. It's excellent . 

CLog has more banding in midtones than WideDr, especially in HD but total dynamic range captured is the same.

Consider 200mbps rather than 300 in 4K to give longer recording times. The quality difference is not huge.

Will check out cinema eos based on recommendations here.

59 minutes ago, Arikhan said:

 

In my eyes detail, contrast, etc. is in Sony much more better. Looking at clouds definition in the sky, there is a considerable difference in dynamic range too. Other opinions?

Yes, the Sony looks a lot better in that vid. But, something fishy is going on here or the user hasn't a clue what they're doing. I don't think the XC10 was set to its best picture style for that test. It looks like it could be the most contrasty one, which has much less DR than C-Log or WideDR. Also, there's something weird happening with compression that is destroying detail in the Canon but leaving the Sony untouched! There's an A/B comparison at about 53 seconds in. At 54 seconds the XC10 turns to total mush, but the Sony is fine. Don't trust this comparison.

4 hours ago, kidzrevil said:

I did a low light test with the XC10 and Canon Log

Was this ISO500? Looks good!

20 minutes ago, tugela said:

The XC15 uses the same processor as the XC10.

Implying a firmware fix might be possible

I much prefer the colour rendition in the first clip - the second looks cooler and more green tinted. Have you tried doing a manual white balance on a grey card on both cameras? I'd be interested to see if the colours would match more closely.

5 hours ago, Lamplighter55 said:

 I can pretty much say with certainty its an error in the calculation of how Canon integrate values from one frame to the next when calculating for motion blur. Basically the maths is off and creating a 'rounding error' so the pixel bins either hold an 'over shoot' or 'under shoot' value - so you get this characteristic 'zoning' of luminance values.

Interesting! According to this theory, would you expect ghosting to worsen with ISO, as we have seen?

5 hours ago, Lamplighter55 said:

So what is causing this ... hopefully it's a firmware only issue, in which case the frame to frame integration algorithm needs re-writing, or it is something done in hardware ... which means a fault in the DIGIC DV 5 chip. Canon will need to check in the later case if it is a batch issue or systemic. Let's hope it only requires a firmware update! The 'good' thing is that if it is (as I surmise), then the solution is just a question of maths - but the problem is how that is implemented - at chip level in hardware or a firmware 'bug'.

 

Here's the latest update from Canon:
 

Quote

 

CANON:
Thank you for taking the time to contact us and for providing a copy of your camera’s settings with sample video of the issue. I can confirm this has been passed over to our European Product Specialists and they have been investigating the issue you reported. At present I do not have any further information but if I get any updates I will let you know.

ME:
Thank you. I realise that you may be limited in what you can communicate to me, but do you think there is any point in sending my camera in for repair at this stage in order to fix the issue?

CANON:
In response to your email, I would advise that at present sending your camera in would not resolve the issue. As this is still being investigated please await further updates from us.

 

I guess you could interpret this in a few different ways. I'm going with "they're fixing the firmware" for now.

This is at 3200 ISO. The area within the red box is where the noise lives!

I'm with you on that one. My take home message is: the advantage of 10bit amounts to a 29% increase in resolution or fps or recording time, or some mix of all three.

If you don't need any of the above, stick with 14bit. But if you do, make sure nothing you're interested in lives in the noise floor (which is shown on the RAW histogram as a dithered pattern - stay outta there!).

On 6/11/2016 at 0:21 AM, hyalinejim said:

I lifted the shadows a bit to see into the blacks.

1 hour ago, Taranis said:

Can you also do some DR comparison?

On 6/11/2016 at 0:21 AM, hyalinejim said:

Try 3 pairs of DNGs here:

https://drive.google.com/open?id=0B1exEpCRAfgFNUNVZVRYN0ZkSms

 

One of those pairs is a high DR scene