After reading about all the issues people are having, I guess I need to say that my experience hasn't been the same. While there are a lot of ways people are using and configuring MBL2, I am using it in two configurations. 1. With a second 27" Apple LED display as a cinema desktop preview, and 2: with a JVC production monitor connected via Grass Valley firewire device. The JVC is an SD unit, so it's not ideal but does provide some reference for the color, especially if the output will be delivered on SD DVD.
I just finished grading (for experimental purposes) three clips, all 720p 24PN. I invoked MBL2 directly from the timeline within FCPX. In each case, I experimented with applying 15 to 20 different looks. When I settled on a look, I applied a mask and I then tweaked several parameters. When I finished, I was returned to FCPX. The first time I did this, the infamous beach ball appeared and I became worried. But, I let it spin and after 20 or 30 seconds, it stopped, fcpx displayed my graded video and everything was fine.
From there, I went back and forth into MBL2 and FCPX and never again had any issues during this session. I have posted below a few still frames from the clips I was experimenting with.
Hopefully this information is useful to someone.
This is the original untouched clip from a Panasonic HVX200A.
This is a section of the same clip applying the classic stock emulation preset of MBL2.
This is a subtle telecine effect with the lips masked and saturated. I adjusted gamma lift, film diffusion curves and saturation on the masked portion of the image.
This final picture uses the blockbuster preset but masks the lips to keep them the original tone.
Very interesting, thank you. I can confirm MBL2 seems to work well with us both on our iMac Thunderbolt and our MacPro 8 and 12 core editing systems. When I apply the filter I get a spinning ball for a couple of seconds when the filter is loading, but from there everything is quite smooth. Rendering is not fast, but most third party plugins in FCP10 do not render very fast anyway.
However I know that more people are having problems with MBL2. I don't know if this has anything to do with the video format that is on the timeline or not. We always work with optimized media. Anyhow, glad it does work for you as well.
Thanks for the stills. For what you're aiming at, it works fine, obviously, but I suspect the preview/render discrepancy is still there, just not apparent under these lighting conditions.
Too bad you didn't include a comparison of the preview in Looks vs. FCP viewer. Depending on your monitor set-up, I bet you will see luminance differences. Output blacks get crushed where they're not in Looks preview. Seems to me like that's happening on your clip, too -- shadow below the lips; eyebrows & shadow behind glasses kind of fuse together. I know one can (and sometimes wants) to do this deliberately, but I am specifically using Looks to avoid that, since too many Motion grading presets have this issue anyway and no one seems to care or notice, while Looks applies finetunable curves for contrast etc. But the result once rendered is never what is shown in Looks, and even if I raise expose 2 stops, I can't get rid of the apparent clipping (and also over-saturation in those areas).
Try this little experiment to "prove" (or disprove -- if it's just my system for some curious reason):
1. the same things happens in Motion; further, in Motion:
2. a blank MBL layer has no "luminance-shifting"effect
3. MBL loaded up with effects at neutral settings produces a minuscule change due to some effects, such as levels, exposure, but not others
4. MBL with effects that do change luminance values of the image, especially curves, color ranges, etc., cause an amplification of this adjustment back in Motion.
5. an image thus affected, fed back into another, blank, MBL layer, will appear noticeable darker also in MBL, but not as dark as in Motion.
And here's another image of the blacks issue, with both previews as a screen grab (no edit)
A lot of 3rd party products seem to be having issue with accuracy of color/luminance, might have something to do with FCP X handling of color space? Until this is resolved, I can't see myself doing any useful grading, especially if the original footage is not totally pro (such as from Canon 550D or Panasonic TM900) and therefore already does have some weaknesses right out the gate as far as color, luma and noise is concerned.
By contrast (no pun intended, I think ), I can get perfectly acceptable, predictable results out of Conduit, both if folded into a Motion preset and as a standalone. It's difficult to use as an FCP grading tool right now, since some crucial I/O handles are missing to attach e.g. curve functions - my "must-have" - to Motion sliders, but it works in principle, so it can be done!
I'm not a professional grader at all, am I seeing something that's not there or being way off base somehow?
Thanks for the detailed research you have been doing on this. Yes, your last image (and ones from the previous posts) clearly show some luminance differences, probably common globally to the MBL2-FCPX integration.
Understand that my original post was not commenting on the accuracy of the grading ability of MBL2, but rather the ability of the program to run and roundtrip within FCPX without the severe crashing issues described by others.
For my purposes, I was able to import footage into FCPX, apply the MBL2 filter, then within MBL2 use nearly every coloring tool I tried to manipulate the image, finish the grade and be automatically returned to fcpx with my footage graded. All done easily without any system crashes, sluggish performance, etc.
Additionally, as others have described issues with multiple monitors, external video, etc. I decided to try this and toggle between sending MBL2 to my SD production monitor and then back to a second full screen preview monitor. All of this testing in the view of seeing just how the MBL2 application performed. And everything works perfectly with respect to the application running as claimed.
The bigger picture, which you brought to light a while back, is the image differences one might encounter between previewing the grading process and rendering the final output. I have done some testing, not to the extent you may have, and I also notice differences. But it seems that reasonably predictable results and some satisfactory comparisons can, in fact, be achieved, but not always. I am not sure if this is common to certain presets, or to what degree adjustments are made, or to certain types of footage used, or if this varies based on the users hardware.
For my intentions, (mostly Corporate, Government, Training, and some shorts/indie productions), I think I can manage to use this program with the hopes that future updates of both FCPX and MBL2 might improve the issues you identify. For anyone producing a major feature film, surely this would have limitations, to say the least.
For comparisons, I have taken some screen shots of the original image untouched in FCPX and also untouched as previewed in MBL2. The shadows here look pretty close to me and I believe would suffice for most work I would do.
Secondly, I have sent the clip to MBL2 and applied a warm haze filter to it. I have comparisons of that in MBL2 and as rendered out in FCPX. Here you can see some luminance difference, but again, for me I could make that work (even if I had to slightly desaturate or saturate the coloring).
As far as the original shadows you were referring to under the lips and around the glasses, recognize that I was not grading that for any specific purpose. I simply wanted to apply some correction to test the application (as described above). I quickly graded all those images and posted them within about 15 minutes.
So here are the latest comparisons: I hope these help to further the discussion
The original untouched image in the FCPX viewer:
The original untouched image in MBL2 on the Cinema Preview Display
(This is a section of the full screen 1920 x 1080 image)
This is the MBL2 Cinema Preview of the warm haze effect.
This is the FCPX final rendered output of the clip as sent back from MBL2 with the warm haze
Thanks so much for that feedback. I'm sort of with you, that I think I can use it (and have used it) for certain grading needs, and in hoping/trusting that these "birthing" issues will get resolved in future. So I'm tending now to keeping my license (which I did get for 50% off, which might not happen again); at the same time, I'm trying to build some Conduit effects for myself for those occasions when I can't get the result I should on Looks. I totally agree that Looks can be useful & used right now, I just wish that developers were more upfront about the issues -- you wouldn't want to buy a car where the braks only work 95% of the time, at least, I'd like to be told! Thanks again! Cheers.
Im just now looking at purchasing MBL2 and am currently experimenting with the trial version 2.0.8. I was wondering if you are still having these issues with the latest version of FCPX 10.0.5 and MBL2.0.8? I'm running OS 10.8 Mountain Lion.