Probably my single biggest complaint when working on manual roto in Mocha is the spline accuracy at varying levels of zoom.
I work along at .5:1 or even 1:1 and when I zoom past 1:1, the spline isn’t really where I think I put it. So I’m forced to waste a bunch of time zoomed in really far checking my edges.
I realize its far from an easy task and isn’t that sexy of a feature for sales, but being able to roto shots zoomed out is a HUGE time saver.
T
Are you using the Zoom window and the Shift+CMD option to slow the movement of the splint point for more accurate placement?
One thing we could look at adding to the zoom window is the actual spline itself, rather than just the axis of the control point, as the curve between points is not always going to be resting where the point is located.
not really. I’ve never found the zoom very helpful for that reason.
and also because I was never certain if it was showing me where the point was or where the mouse cursor was when I clicked… in other words, I wasn’t sure if “snapped” to the location of the point I clicked on, or just showed me the relative distance the cursor was at the moment I clicked.
I could certainly train myself to use the zoom window if it had the spline in it. It’d be nice if it was somewhat numeric, too. As in: ‘show me 2x’s the pixel resolution.’ it doesn’t seem to be very smooth when I try to zoom it with the “Z” and mouse.
T
Thanks, we’ll keep this in mind.
How would you like spline control to work in varying levels of zoom outside of using the zoom window? Is the spline getting in the way of seeing an accurate placement? Do you find the edges are not scaling as you expect?
mostly the edges seem to not be in the same spot relative to where the visual cues I use to lay down the spline are.
Splines seem a touch more intrusive than Shake. Honestly always did 90% of my manual roto’n in Shake. If I though something would benefit from tracking I would export the track into shake and roto through the track.
I was forced to upgrade my OS and lost the use of Shake at the same time The Foundry gave me a deal on Nuke. Then suddenly Shake started working again, but wow, it does NOT handle 5k at all compared to Nuke.
I like Mocha better than Nuke for roto, especially how the spline key frames are carried over, so now I’m doing the bulk of my rotoing in Mocha.
There’s really only a few hiccups still that cause me to click more than I want. Aside from wanting to be able to roto zoomed out, the base behavior of how Mocha decides which knots I’m trying to select with my clicking, could be more intuitive. I find I have to repeat attempts to select some points, and sometimes failing to select a knot will move the last selected knot, even though I’m on the other side of the screen.
Overlapping shapes on the same layer could be better, too. Maybe if the target click zone for a knot shrunk if there is another knot nearby. Sometimes I can’t click directly on a knot if another spline is nearby. I have to drag select.
Nothing as beneficial as rock solid accuracy on the first pass. I’d like to feel confident enough in my spline placement on the first pass to actual skip a QA pass and deliver it.
T
this could be a double post:
I can’t stress the importance of being able to select more than 1 but less than all 4 of the Surface Points in the AdjustTrack Module.
I constantly need to move 2 points together.
OR.
If the square/box manipulator would work on the 4 points, That’d be really nice, too.
Soft Select!
Soft select for knots AND soft select for Uberkey.
Maya-like select and deselect hot keys would be nice. (Shift + Click reverses selection, Ctrl+Click subtracts, Alt/Cmd + Click adds) As well as adding selecting/deselecting to the undo cue.
Saving various selections with their unique pivots would be a great feature, too. Almost like a bone/skin structure.
Seems like a new bug in mocha 5, but I’m having a heck of a time using shapes to hold out other shapes for tracking. Seems to rarely work any more. basically I’m key framing shapes to avoid non-participating features that occlude. Almost seems like it could be a caching issue. Not sure.
Love the GPU tracking on CentOS with a 1070 nvidia card, btw! wow. smoking fast even at 4k.
T
set the in/out for all selected layers.
seems to only work for one at a time.
T
copy/paste keyframe of shape added to Right+Click floating menu.
ProRez 4:4:4:4 codec on Linux
Nuke does it. Blender does it.
We are watching this thread and recording requests. 
We can’t guarantee anything at this stage, but it’s very useful to read your thoughts on the functionality going forward.
yes!!! ++1
Spatial and temporal soft-selection would be very useful
Robert
My personal #1 annoyance …
I hate hate hate, (there I said it three times so it’s a Trumpism and must be really, really, really true), I hate the way Mocha locks you out of the UI as soon as you press play .
I can’t think of another vfx app that does this - all of them allow you to click anywhere in the time slider and either stop and jump or jump to the clicked point.
To double down on this “unwanted” behaviour of the UI you have separate buttons for many functions including Play & Stop that constantly require me to look away from the feature I’m trying to focus on in order to press the right button. When you rely on caffeine the way I do this is no small feat.
Why not use the spacebar you say ? I’m a Nuke user and that’s whole other request / whine I will document.
My personal #2 annoyance …
Closely following on in the same behaviour - when I click in the time slider to jump to a frame I cannot simply hold down the mouse/pen and retain control of mocha’s position in time - No I have to release the mouse, move it around a fraction to activate the select and drag function and then drag to another point in time and then repeat, repeat as infinitum - why ? Show me another piece of software in this arena that behaves this way and I’ll buy some coastal real estate in Key Largo from you …
Hey, thanks for using the new forum!
We freeze most of the GUI during playback so that you don’t inadvertently generate keyframes, but I can see the need to allow control of the timeline in this instance.
So, here are some things in this thread that we currently have scheduled for v6:
Note that “scheduled” does not necessarily mean “will make it”, as we have some major features incoming that will take priority.
We are looking into the soft select and placement accuracy. These require more investigation. Clip codec management in v6 will most likely change, so we’ll have to see about ProRes support on Linux.
Tracks could ‘fail’ more gracefully.
for example, if a surface corner leaves the screen, more emphasis would be placed on Translate / Scale than Rotate / Shear so noise and distortion is reduced in favor of accuracy.
Also, I’d REALLY like to have Rotate before Scale in the transform matrix.
Most times when I can’t get a usable enough track, I can get a fairly usable translate/rotate track out of Nuke.
I will even average the 1 / 4 and 2 / 3 corners and use those 2 new points to drive a translate rotate matchmove.
And finally, the Holy Grail of features:
Head Recognition.
yeah.
as in “Hey this is a face. This is the pivot. This is the rough shape.”
Combined with camera lens assumptions, graceful tracking failure, and some manual tracking, seems like you could get a reasonable head position rotation ‘track’ and use it to place hair, ears, glasses…
IDK. Just something I deal with a lot.
Far more useful than trying to get Mocha to do full 3d solves, anyway.
T
Replicate “Export Shape Data” functionality in “Export Tracking Data” dialog.
Currently in Mocha Pro standalone or OFX for Projects where I have multiple Track Layers which I need to bring back into Nuke I’m forced into multiple round trip’s Copy, Save, Quit, Paste, Relaunch, Repeat etc or Exporting Multiple times to discrete Track.nk files somewhere on my system for each Track Layer - which all have to be renamed to match the internal name’s so I can keep track of what is what.
If that’s a long sentence to read imagine how it is to actually do it …
… here’s the thing though…
For the export to desktop workflow at least all the functionality I need is already in the “Export Shape Data” - I can export the “Selected Layers / All Visible / All Layers” and they end up on my desktop with the correct name and extension ready to come back in !!
At a minimum can we replicate the “Export Shape Data” workflow in “Export Track Data” ?
Thx
It’s entirely possible to do this Andrew, although it would be limited to the hosts that could support it.
We have been planning to overhaul the export controls for some time now, but this may be a good first step.
Of course if you use the standalone you could implement this functionality in the Python API instead of waiting for us, but we’ll what we could do.