[Petter Stahre]

[Alex Ojeda] “Ive been doing some tests, and I think youre right about the generators. The export takes about 10 minutes to export 44% (at the beggining of the clip I have a 3d shadows generator), but the it really slows down, and the next 20 percent takes about an hour and a half. In the middle of the clip I have a generator with drop zones, and a motion 5 overlay I made stacked up on top of it. “

Just to share a similar experience: I’ve just ended a Motion project which had the exact same problems due to to a particle emitter I used (and 3D). The beginning of the 20 seconds movie was rendered fast (still some 20-30 minutes), the last 5% took hours. And in the end I had to render parts of the project in to two different .mov-files in order not to get a quicktime render error (same happened no matter if Motion or Compressor did the job). The first 95% of the project was rendered with motion blur, but the last 5% I had to render without motion blur since the particle emitter had grown to much. (Luckily I didn’t need motion blur on that part.)

As mentioned in the thread, rendering from Motion to a .mov and then import it into FCPX could help a lot if possible.

[Alex Ojeda] “One last thing ive noticed, and cant get rid of it. I add an opener generator to the beggining of my timeline, it has 5 footage drop zones and a logo drop zone and 3d shadows. It renders, I preview it, and it looks good. No problem. Then I add next to it a 20 seconds footage clip, and a cross dissolve transition. It renders, i preview it, and now the footage on the drop zone do this weird flash, kind of like a stutter, and the logo shifts its position for a half a second, making it look weird. I though this had to do with the computer not being able to preview it as it should, but I exported, and those behaviors exported as well to the output .mov clip. I changed the transition, and the same happens. I remove transition and clip, and all well again. “

Sounds a bit like the bug that was corrected in FCPX 10.0.8 and Motion 5.0.7 although I only had the problem with random single frames (not half a second of rendering as you mention). Are you using the latest Motion (5.0.7)? If so, have you tried opening the generators in Motion since the upgrade and saved them again? This is just a thought – I’ve got no idea if should be needed.

Best,
Petter

[Petter Stahre]

I don’t know if this is possible in Keylight but if not, here is an alternative using Phyx Keyer which can be bought for AE (199$):

https://www.studiostahre.se/temp/GStest1080p.mov

Just tell me when you want me to erase the link.

What I did was this:
– basic grading (sorry, a bit yellow 🙂
– since I suspect you don’t use camera movement for the shot I created a still image (screenshot after grading) without the singer, using Photoshop … which is a poor mans version of filming the empty green screen scene which of course would have been better. This was my quick-and-dirty version: test
– I let Phyx ScreenCorrector use that image to analyze differences between your shoot and the empty scene
– I then applied the Phyx Keyer and made a copy of the layer – one layer had settings for the upper body and one layer for the legs and feet.

I’m not satisfied with the shadows, they could benefit from other settings or a third layer, but for the test I stopped there since it’s also a matter of how you’re going to use the singer.

I would not be surprised if the same method could be applied in Keylight !?

[Petter Stahre]

Since this thread got me curious and it was a long time ago since I used C-log for green screen I decided to do a little test.

Since I’m alone in the studio I filmed a mannequin while shaking it 🙂 using my preferred custom profile and then C-log (but slightly edited to get rid of some green tint).

I then used FCPX and Phyx Keyer 4. And now I seem to remember why I stopped using C-log for green screen work … it’s much easier getting a clean key if I use my custom profile, at least in my workflow. I’ve done some basic grading to the C-log shot but still I can’t extract as good key, it’s quite noisy if I don’t use quite harsh settings on the matte.

My custom profile use EOS Standard as a base (which is awful by default, yes I know!) but I’ve tweaked it a little and it works for my kind of work.

In the video I’ve scaled the C300 shoots to 200% which means when you look at it in 960×540 pixels (which is the size of my player) then it should be a 100% crop of the 1920×1080 original.

I think if you pause the video you can see that the software has trouble with the edges of the blurry hair when using C-log. It’s also moire noise in the fine hair just over the top of the head in the C-log shot.

All is shoot directly to CF-cards (50 Mbits/s). From what I understand an external recorder and ProRes results in less compression artifacts but the color data will be exactly the same (4:2:2 8-bit).

https://www.studiostahre.se/tester/spelare.asp?film=c300greenscreen.mp4&bredd=960

No science … just a quick test.

One of Canons/Sam Nicholsons launch movies “XXIT” shot with the C300 relies heavily on green screen from what I know. What software and profile they used I don’t know but I would still guess C-log since that is what Canon promotes.

[Petter Stahre]

While I don’t use AE I use my C300 extensively for green screen shoots and has had no problems which I can relate to bad quality from the camera.

I use FCPX together with the built in Keyer and now more often Phyx Keyer suite since release 4.

I find that Phyx Keyer is generally better even though it has less options for fine-tuning than the built in Keyer. I shoot simple corporate videos where I shoot the background and the subject at different locations. Quite straight forward and not much motion blur.

As for the C300 settings I’ve been using both C-log and now more often a custom made profile based on the EOS Standard profile but without sharpening, lighter shadows and color corrected to get rid of the slight greenish tint that I think C300 always has. Since I don’t do heavy grading I found that a ready-baked profile suits my workflow better.

And I’ve never used any external recorder. But for my work this has produced very good images.

[Petter Stahre]

… or it’s actually a feature 😉

If you have 5 clips from the same take and have made slight adjustments to crop/size then the masks would be applied at the wrong place if it worked otherwise.

It would be neat though if there was a choice to paste the attributes relative or not.

To make FCPX work the way you want you could first position/resize/crop all the clips. Then make a compound clip of each.
Then make the color corrections and masks to the first compound clip, copy and paste to the other compound clips. Since the compound clip itself isn’t rezied or cropped you should have no problems with the mask.

[Petter Stahre]

Thanks a lot Todd.

Yes, I’ve managed to assign ISO to a button but if I assign shutter speed to the other possible button then I loose iris control. And since I use EF lenses with no manual control of iris it doesn’t help me much. Then I have to use the joystick for iris control instead.

So I’ve resigned to use the joystick for all three settings. I tend to forget I have a button for ISO 🙂
It works, but Canon could have solved this much better. I’m still hoping for a firmware update.

[Petter Stahre]

While I agree on almost everything Todd writes (great reply) I’d like to add some thoughts.

I do corporate videos and bought the C300 some 6 months ago. I have no experience of the Scarlet but considered it as an option. If I bought a new camera today I would have serious problems in deciding which one of them to buy since REDs lower price and the forthcoming Dragon.

As a customer I also feel annoyed that Canon charges quite a big premium for what they deliver, considering the new Sonys and REDs lower price.

(And by the way, here’s some great information, specially reply #7: https://www.dvinfo.net/forum/new-canon-cinema-eos-camera-systems/502822-reasons-go-c300-over-red-scarlet-x.html )

Mixed thoughts on C300:
+ Very reliable. I haven’t had any problems with it so far.
+ Base ISO of 800 and very good high ISO images. (While I do think grain is visible from 3200 ISO it’s not a problem, and as many says quite “organic”.)
+ Good battery life.
+ It has a fan but is quite silent. I’ve had no problems when recording audio.
+ CF cards are cheap.
– The monitor is ok but can’t be used to judge white balance, specially not when using warm light. Side by side to a better monitor there is a big difference. Also the color space is very limited. Saturated colors gets flat very fast.
– The idea of having to attach the monitor to be able to use XLR is bad. It would be nice if Canon developed some smart accessories (like a small XRL-only accessory).
– Ergonomics could be better (but no problem). Why no easy way to adjust shutter speed and ISO. You have to use a tiny joystick for this. Three buttons for iris, shutter speed and ISO would be fine.
– The roll off to overexposed could be softer.
– While DR is ok I’m not impressed.

Mixed thoughts on Scarlet (while I don’t own one!):
+ RAW. I really miss RAW, to be able to adjust WB afterwards and slight exposure errors is for me a big bonus.
+ PL and EF lenses can be used. Changing mounts can be done by yourself (even if it includes using a screw driver).
+ Upgradeable to Dragon-sensor (with 17+ stops of DR?), but supposedly at a price. (And from what I understand you’ll get an Epic Dragon in exchange, so there really is no “upgrade” of the Scarlet – but a substantial rebate for an Epic Dragon update, if I got it right.)
+ DR might be better than C300 (13,5 stops vs 12 stops).
+ 10 bits 4:4:4 vs 8 bits 4:4:2.
+ 4K. While I never deliver more than 1080p i really appreciate to be able to crop and zoom in. For a 1080p production this can be done quite much without loosing sharpness. This one I really miss in C300!
– Base ISO of 320 (but is supposed to look good at ISO 800). I suspect I would need more light in general if I used Scarlet. I also would suspect high ISOs not to be as good as the C300. Someone’s got experience?
– / + Workflow might take more time (RAW conversion) and lots of more storage. On the other hand you can buy an Atomos Samurai and record ProRes. You don’t have to record in Raw! I think this is a point missing in many debates. Atomos Samurai costs $1600 and then you can use cheaper (than Red) SSDs.
– Fan noise is supposed to be quite high. Comments anyone?
– Battery life is supposed to be bad.
– Media is very expensive (RED SSDs).
? Software reliability? I’ve read recent threads that some functionality still isn’t implemented and that there have been different bugs since release. Note: this is only what I’ve read and it can be wrong. Maybe an owner can chime in?

I’m writing in a hurry, sorry for misspellings/bad grammar. Now back to work…

[Petter Stahre]

I tried to repeat your problem, creating a color correction with 4 masks and pasted the attributes to another clip, as you do in your movie.

It works fine for me, but I noticed that if I change the scale of the other clip and then pasted the attributes then the masks where out for place.

So it seems that the position of the masks is relative to the scale (and maybe other parameters under Transform / Crop / Distort, but I haven’t experimented more). Maybe you can find your problem there?

I tested to Paste attributes and checking both “Color” and “Transform” (all boxes) and then the mask got applied correct. But of course that changed the scale of the second clip and you will have to manually change that back afterwards. (Or duplicate the clip before pasting the masks, and then copy attributes from the duplicated clip and only paste the Transform-parameters back to the original clip. Then you delete the copy. Maybe that could work, even if it’s awkward?).

[Petter Stahre]

Thank’s I’ve been reading about it but not tried it.

Now I’ve tried it, and it seems good. The watch folder-function is a deal breaker (not having to manually add what projects to keep track of).

It also answers my question if backing up the project file itself is enough – it is.

Still, I’m interested in going the Automator-way for different reasons, so if someone knows if it would be possible the way I described I would appreciate thoughts/hints.

[Petter Stahre]

While I don’t know what causes the bad usage of your CPU I have found a way to use Compressor that is much faster (if using Compressor is an option for you?).

I first read about this here on the Cow and think this might be The Holy Grail of compression workflow. I started using it after I was annoyed with how long compression took when using the Keyer and other FCPX-effects.

Workflow: Send your job to Compressor, set up an export to ProRes, chose “Job/New Job With Target Output” and then choose the final format&compression you would like.

This means a ProRes is first created, and then Compressor is using it to calculate the next (final) job.

For me this has cut compression times with a factor between 3-6x, when the final format is using H.264.

It seems like if Compressor has to much to think about at a time everything goes slow instead 🙂 But if you let it concentrate on first building a proper full sized ProRes, and then compress&downsize&… it gets much faster.

Sorry if this isn’t a workflow you can use (or if you already know about it), it just made my life so much easier when I started using it I had to share 🙂

(I still would expect Magic Bullet effects to take some time to render though.)

Best,
Petter