hyalinejim

I don't think it's that bad. Are you referring to the screengrab above, or the DNGs I posted?

Just copy the contents of your card to your computer as a backup so you can go back to your current setup later. Then delete them and stick on the new stuff.

Links are here: http://www.magiclantern.fm/forum/index.php?topic=5601.msg174306#msg174306 You're looking for: raw_video_10bit_12bit_2016Nov04.zip Previous experience with ML will be necessary. Use the raw-rec module rather than mlv_rec (you still get MLVs) I'd take that with a pinch of salt. 2880x1206 (Cinemascope ratio) @ 24fps 10bit is 99.3MB/s and I think the 5D3 tops out at 100MB/s no matter how fast your card is. Also max vertical resolution is 1320, so anything higher than 2340 can no longer be 16:9 ratio FYI. I think if they were going to panic they would've panicked when ML brought full HD 14bit RAW to the Mk3 and smaller resolutions to countless other EOS cams. This new development is welcome, but to put it in context it basically gives a boost of 29% to achievable frame rates or resolutions for existing ML cameras. ML has a steep learning curve when it comes to installation and workflow, and requires an investment in fast high capacity cards and storage, but once you have it nailed you're golden. I find it very difficult to be impressed by compressed footage from any cam after shooting 5D3 MLRAW for the last year.

Nice colours too

It's looking damn good to me. The IBIS when he gets on the little train is terrific.

Good stuff! I guess EOS standard has less DR so easier to compress. AFAIR you can dial down sharpness to zero in WideDR.

The XC10's 1080p is quite compressed and has a lot of artifacts. Its 4K downscaled to 1080, however, is lovely and stands up to moderate grading quite well. But of course, the motion ghosting that we've been going on about here a bit lately precludes it as a serious contender for image quality.

Thanks for this

Here's a comparison at ISO 100, exposed to protect highlights, 2560x1072 (14 bit gives 28 seconds, 10 bit is continuous) in greyscale crop mode. I lifted the shadows a bit to see into the blacks. 14 bit is top left, 10 bit is bottom right. Try 3 pairs of DNGs here: https://drive.google.com/open?id=0B1exEpCRAfgFNUNVZVRYN0ZkSms I think that 10bit will work really well, as long as there's nothing you need to pull out of the noise floor. Remember that the 2K DCI spec for Cinemascope is 2048x858... which is 80% of the above res. This allows for a significant amount of re-composing / stabilising / downscaling for increased resolution. Interestingly, both types of DNG are the same size in bytes - don't know what's going on there.

@kidzrevil I know you need a fast turnaround on files but tbh you would love the colours out of 5D3 ML. And no ghosting

That is awesome news! Got one for my XC10, but I'll be glad to stick my 5D3 on there.

It would help a lot with bitrates for sure.

Yes, but any gains in resolution are only offered by the crop mode with the stuttery greyscale preview - it's off center and quite difficult to follow live action. The normal mode and 3x crop fast preview mode are limited to 1920x1080 AFAIK.

Canon appear to be looking into the issue at the moment. They've asked me to forward the settings for my camera. I've sent this on to them now and mentioned that I've tried various combinations of settings but all produce ghosting on my unit. Have any of you guys had requests for further details like this?

I like pond5. Can't compare them to musicbed as I haven't used that site. But I've found a range of good stuff on pond5 after a little bit of clever searching. I like to search for a specific artist that I want something to sound like, rather than general tags which are overused. For example, I've searched for "Bon Iver" rather than "folk", with good results. Also, has anyone checked out this: http://www.dvinfo.net/article/post/first-look-music-vine-a-fresh-new-music-licensing-platform-for-filmmakers.html

I agree. All of the seven or eight XC10 ghosting tests that I've seen have shown it, but it doesn't mean all XC10s have the problem. If anyone has a clean XC10 please post footage. And the one clean XC15 test doesn't mean all XC15s are free from the problem. If anyone has a ghosting XC15 please post. If you're thinking about getting either model I would strongly recommend you try before you buy, or look into your seller's return policy.

If it is hardware, they should be able to fix it by replacing the faulty components. It would be nice if it were a hardware problem that only affects our units. But we've yet to see output from a unit that doesn't have it. Otherwise I would've sent mine in last week .

+1 Mp4 looks just like MXF to me. No better or worse.

- my emphasis. I agree 100% @kidzrevil. Also when shooting at ISO 1000 in a camera which Canon says offers.. Anyway, back to some more testing. This time - what is the difference between high and low bitrate 4K, when the final delivery is downsized to HD: The higher bitrate is a BIT better in this stress test. However, for me I think I'd be happy to shoot at the lower in order to get more shooting time per card. Or if you're shooting at 305 and you're worried that you're gonna run out of space, switch to 200 to extend your recording time.

@mercer The answer to the first question is: those are the most significant differences, as well as the fact that larger sensors generally have less noise. The answer to the second question is: if the focal lengths and apertures are equivalent then the field of view and depth of field will match and the aps-c will not stand out. But the lens that will match the desired focal length and depth of field may not exist for aps-c or it may be more expensive than you'd like. The take home message is: Use the laws of equivalence to decide if a given format's lens selection will meet your desires and budget for: Angle of view - how wide or how telephoto you need to go Depth of field - how little or much of the image do you want to be in focus Light gathering ability - your preferences for depth of field in a given lighting situation with regard to the sensor's noise performance This will allow you to choose the best format for you, or to match one format to another within the limits of available hardware.

They told me this last week: "We have passed your request onto our European Product Specialists for further investigation and they will analyse the video files you provided to see if there is an issue. We will contact you again shortly after the video has been analysed." So I'm expecting feedback on this from Canon. I'll be shocked if they say there isn't a problem. We think the XC15 is clean, from Tom's footage, so I believe the problem is fixable - isn't the XC15 essentially an XC10 with slightly different firmware?

Contact them and ask them. Say exactly what you've said here - that you find the image compromised and are reluctant to use it for professional work. It's in their pro line of camcorders and is touted by them for its good low light performance. They will fix it, I think, but the more people contact them, the better chance it becomes a priority - and sooner rather than later.

Well, yes, of course because the aperture is not equivalent even though the focal length is. When comparing sensor sizes you multiply the focal length and aperture of the smaller sensor camera to give the equivalent focal length and aperture of a full frame camera that would give an equivalent angle of view and depth of field from the same viewpoint. I honestly believe that this topic is so complex that it is quite difficult to hold together in your head the relationships of the variables in such a way that corresponds to reality. Hence, extended and heated discussions like this one, and the 2000+ comments on the dpreview thread which, as previously mentioned, is worth a read and a re-read.

FilmConvert Velvia - all of the stocks seem to have various colour casts so I have to wiggle white balance and red-green tint prior to FilmConvert.

I think so - I had read that perfect exposure was really important with CLog and WideDR and that if you overexpose then highlight colours get washed out. But it seems fine to me to ETTR based on my tests so far. And 500 ISO does have noisy shadows so it's a good way to avoid that. Remember that the signal goes to 109 IRE so when you see 100 zebras there's still info there in the superwhites.