Jump to content


  • Posts

  • Joined

  • Last visited

  • Days Won


Everything posted by jan1

  1. I followed the tutorial above and the examples by Yoav and Cristobal, but still have one thing that didn't work as I expected. I was doing skin cleanup with the clone brush on a face. I needed a total of 5 shapes. I also did the motion tracking with one point in the tracking tab of the vector paint effect. The problem I ran into is if I do 'Current Frame' and 'Selected Shapes' the 'Copy Tracks' button remains grayed out. After some experimentation I found that this only works if I select one shape at a time. Then it's perfect. But if you have quite a few shapes that is extra work. Is it a bug or is there a reason why a track cannot be applied to multiple shapes in one operation?
  2. Hi @cbolanos@sgo.es Spent some more time playing with the Elements mapping. It's still very much work in progress. Current mapping file attached. In doing so, I found it easy enough to change the mapping with the Tangent Mapper. But I also found the current Mistika mapping library extremely messy. I think there was some original idea on how to organize it, and then stuff was added in semi-random places. It also doesn't look like it's complete. It would be nice to organize the folders by user interface section (mirroring where things are found in the interface / by mode). And the library changes based by mode, so sometimes it's not clear where to find things. Some mistakes - the SB 'show more' has a blank label, just able to find it because it was next to 'show less'. There are others like DefaultFx which are all blank. Things I couldn't find: 'add keyframe' and more importantly some short cuts to frequent fx, such as color. I would be great to stay in edit mode and have a key on the panel to add a color node without having to switch back/forth to FX. Unless there is another trick? I could easily see one of the button panels be populated in the second bank with your 9 most frequent FX nodes. Also, are the entries for the 'Soft Clip' in Primary. I'd love to map the 4th trackball to softclip when it's not used for Jog. A few things about the changes I made. Above I mentioned the keyer on the trackballs. I swapped the history and library buttons to make history available without bank switch. Moved a few of the SB buttons around on the Mf buttons bank to place them all next to each other. That freed up some fields in the 2nd bank of the modes buttons to configure SB/Lib and keep the keyframes. Overall trying to minimize having to switch banks as much in the way I intuitively work. The default mapping is good, but I find myself having to switch too much. Hope this helps... AllKlier R1.xml
  3. Working on another complex grade. Ran into an issue with template groups. They work really well if you keep your viewing focus on the template instance (top most eval tree layer) and then move your focus up/down to adjust certain elements. In this case I have a key further down in the eval tree, and that part of the tree doesn't feed into the output (it's meant as a separate key branch to only generate keys). Thus I cannot see the key matte in highlight mode. I would have to double click on the keyer layer to set both viewing and control focus there. That works in a regular eval tree, but if that part of the tree is part of a template group it only shows the placeholder, the template doesn't instantiate. Is there any work-around? Is there a reason that the template cannot display the actual images instead of the dummy? First screenshot is when the visual focus is further down, second screenshot is if the visual focus is at the top.
  4. Ah, yes, that's exactly what I was looking for. I guess it's called the 'Link' layer. And it has exactly one argument, though it's not very intuitive, as it is the 'source offset' in the tree, rather than something more descriptive like a layer name, etc. And I don't know if it can be created in the timeline editor or if you always have to go to the node editor to do it?
  5. I was looking at the Feedback FX before, but wasn't totally sure what it does. I just played with it, and it kind of works. But sometimes when adding effects it wants to collapse the stack. The left is what I think you meant with Feedback, you put it once above the clip to 'write feedback' and then you can reference feedback in the tree above with scope set as such that those have 0 inputs. On the right is what I see others do in tutorials, just copy the clip multiple times. Works fine, just seems non-intuitive from a tree perspective. In my mind the tree should have a single input that can be referenced multiple times. In Resolve if you create a compound node, you can put a null node as the first node in the tree to act as a source splitter. Or you can have multiple inputs.
  6. Hi Cristobal, Thanks. To clarify Q1: As in this example it can be done with a dummy node. But in my understanding dummy nodes only work in a template group. What if you wanted do the same in a regular node tree before or without turning it into a template? Also, even if I create a template group, when nesting them for a complex tree, I have to keep adding dummy nodes at each level to propagate it, which made the tree at least several nodes taller. When watching tutorials (like the frequency separation) I see a lot of folks just duplicating the clip. Maybe there is no downside to just doing that (as it's just a reference to the original file), and I'm overcomplicating things ? It would just seem cleaner. Also then you could just copy a tree to the next clips left/right, without having to worry about duplicating the correct clip, since everything sitting above the clip is clip neutral. A different way of looking at it - in your green screen tree above you have a clip 009M9 twice as a leave. If I understand correctly, the only way I can have the main source clip once in the tree, unless you copy it, correct? If you had a 'reference' node, then you could have the source once, and all other leaves in the tree can simply reference the first one. Kind of the same way you can reference other vectors or inputs for keys. Such a reference node would have a single parameter, which would be a drop down (similar to the input drop down), and instead it would list all source footage clips that are in the tree already.
  7. jan1


    I don't use Lightspace CMS. What would that 3D Lut do? Final output mapping or scene referred color? In that case just create a display filter with a 3DLut layer and assign it to live (and GUI if appropriate).
  8. Hi Christobal, Happy to share that mapping with you, and I love the idea of having a few alternate mappings available for download. Let me play with the mapping a bit more. There are a few other things I found awkward in the default mapping, but not nearly as maddening as the key inputs. Which BTW, I would keep the mapping on the knobs for the keyer, I was just using the trackballs as an alternate input since they were unused in the default mapping in the keyer state. And you're correct, Resolve does not allow for modifications of the mapping which is a major drag, because their default mapping is quite poor.
  9. Thanks. Sometimes things hide in plain sight. There I was trying to drag the vectors with all combinations of shift, cmd, opt....
  10. I was playing around with the Tangent Mapper for some custom mappings. I think I read in another thread that SGO is considering a new layout. The one thing that bugs me about the default layout is that for the keyer it's hard to setup the initial values. Specifically with Lum/Sat defaulting to a narrow band and being on separate pages it's too slow to dial in a key. At the same time I love how Mistika uses the track balls in different context to speed things up, like window size and position. So here is my new Tangent mapping for keyer inputs on the trackballs, extending on the existing Hue Angle. Now it's super fast to setup a new key: Left Trackball: Ring: Hue, L Soft X: Lum Lo Y: Lum Hi Alt-X: Lum Soft Lo Alt-Y: Lum Soft Hi Center Trackball: Ring: Hue Angle (existing) X: Hue Range Right Trackball: Ring: Hue, R soft X: Sat Lo Y: Sat Hi Alt-X: Sat Soft Lo Alt-Y: Sat Soft Hi
  11. I spent some time replicating a recent beauty grade I did in Resolve in Mistika to make sure I understand the corresponding tools. Along the I learned a lot of details, but also came across a few things I'm missing. Several of these may exist, and I may just know exactly where they're hiding. Anyway, attached is the node graph from Resolve. It's a pretty typical graph that is replicated on each clip and then adjusted as needed. It largely divides the grade into two branches, one for background and one for skin which merge at the end. And also, all skin keys are sourced from the original balance node to avoid any cascading effects. I did manage to create a template group in Mistika that more or less does the same thing. Attached a screenshot of the eval tree and the vector stack of the main skin color node. I think this could be a helpful set of tutorials over time for many new users coming from Resolve on how to translate some common node graph elements. Here are the questions and observations: In my eval tree I want to make sure I'm not copying the original clip, but rather can reference it in different parts. I eventually figured out how to use a template group with dummy1 nodes to achieve that. But I was wondering if there is a different mechanism outside of a template group that allows any layer in the eval tree to be used by reference elsewhere in the same tree. The same way you can select the external key resources, it would be great to have like a 'reference' effect that has a single parameter, which is to select another eval tree layer as source. I figured out a way to replicate a key mixer in Mistika. You use a vector per key and then use the mask/add/sub operators in the external key to combine them. The one issue I ran into that, is that when there's a window on the vector, that vector affects the external key input as well, which can be a problem. It makes sense for windows to affect the internal key, but affecting the external key should be at least optional. Is there a way of re-ordering vectors in a color grade node? Once in a while as building a complex color node, especially with complex keys it may be necessary to insert a vector higher up. When referencing external key sources, it would be helpful to get the names of the vectors rather than V2, V3, V4... I love that in the pre-key window you can select a different source for the key. That makes it so much easier to make sure the key is un-alterated.
  12. Hi Cristobal, Thanks for the quick response. Yes, the ticket response helped answer the question and everything is working now. So to summarize for the community. If working on footage larger than your external display (e.g. 5K) that needs to be maintained at full resolution for processing and rendering, using the display filter for 'Live' only to scale the timeline to any external displays is the correct approach. However, the display filter cannot be applied to GUI, otherwise any shapes that are drawn will actually be mis-represented on the GUI screen and not match the actual affected area. That's not an issue though since the GUI display can always be resized to fit the screen anyway. Jan
  13. Right, that's what I did. Make sense if you need a single track. In my case I mapped the surround files (6 mono files) to channel 1-6 in L,R,C,LFE,Ls,Rs order. Then mapped the stereo file to channels 7 and 8. But when I render that, I get 1x 8 channel stream instead of 1x 6 channel stream and 1x 2 channel stream.
  14. Yesterday I was creating the final file for film I had previously graded. I needed to render my existing graded output with both a stereo and a surround track that was finished in ProTools. I thought I would get a feel for Mistika in terms of finishing software. I successfully imported all the materials and went to the audio router panel and mapped all the channels. But from playing with it and reading the manual it doesn't appear that Mistika has the necessary infrastructure to understand track types and the meta data for individual channels inside tracks. I can map all 8 channels, but not separate them into two tracks and append the channel order in each track. I saw that the DCP module can handle all this, but not the regular output module. I also saw some notes about the avconfig file where you can manually setup ffmpeg parameters, though that would be quite hands-on. I have other software I can do this with instead. But when I read that Mistika is good as finishing software I thought it would be able to handle this too. How do folks use it create their final deliverables?
  15. I already filed this as a ticket, but since everyone is on holiday, if someone here knows the answer that would be much appreciated. The project at hand is RED 5K RAW transcoded to ProRes 4444. I do need to work on color and some VFX and then render back out at 5K for the editor. Since my video out is only UHD, I setup the project as 5K master format and UHD video resolution. On startup Mistika gives a horizontal resolution mismatch, which is fine. I then added a display filter with a framing node that does as 5K->UHD adjustment. That works great for working on the project, and I was able to do color and all the other things. But once I drew a window in a color node, the actual color adjustment does not follow the shape. It's scaled down as if the resizing transform is applied twice. Doesn't matter if I turn off the display filter. Maybe just a bug, or maybe ignoring the resolution mismatch warning wasn't proper. What's the correct way of setting up a project that has a timeline resolution that has to be different (bigger) than the display resolution?
  16. I was one who originally asked about it. I do think that most people who are primarily using Resolve with external reference monitors will already have BMD I/O hardware. If you are hoping to get a good number of them to adopt Mistika, having support for their existing I/O path will remove a major resistance point. Investing into a second AJA interface is generally at least a $2K investment and not simple plug and play, it would essentially double the price for Mistika for them over an 18 month timeframe. What I don't know if is how many people you are targeting are coming from Resolve vs. other systems, or this is their first serious grading platforms. For me personally, I now have an AJA interface and it works well for me. Would I convert to using Mistika on BMD? Not sure yet.
  17. I think I may have sorted it out. It had to do with the val tree and keeping the focus on one layer and the settings on another layer. Once I option clicked on the grade layer I was able to adjust without losing the wipe. The intricacies of a new UI.
  18. Had a bit of a rough start on my first client project with Mistika, had to go back to Resolve for now. Hoping I can figure this out though. There seems to be an issue with when the display/output refreshes. I've had a few instances where after loading a clip from the media manager and starting to grade on the panel, the screen wouldn't refresh and reflect any changes until I would step one frame left or right. A tad annoying but manageable. But then I hit a bigger problem. This particular job (beauty demonstration video) requires a background replacement to a specific Pantone color. So I created a big swatch in Photoshop and saved it, brought it on a separate track above the main timeline and added a wipe so I could see a stripe of the reference color and then key the background and adjust as needed on the main timeline. Except the wipe would constantly disappear. I'd move the monitor mark left/right and the wipe would be there. As soon as I hit the panel's level trackball the wipe disappears and I would have to adjust blindly. I step one frame left/right the wipe comes back. I try to adjust color the wipe disappears. It was impossible to see them side by side while adjusting the color at the same time. Seems like the playback should more seamless? Could this be related to any performance options in mConfig or something else? (Config: Mac, AJA I/O, Tangent Panel, latest build in 8.8.0)
  19. Double check all your settings in mConfig. What through me for a loop at first is that there are two checkboxes that have to be clicked to enable the panel (for Elements, Wave may be similar). One is 'enable USB surfaces' and the other one is the specific panel model. The UI is a bit confusing.
  20. jan1

    Spanned Media

    Tracked it down and filed ticket. Turns out it wasn't code specific, but for some reason in this particular project Avid included additional audio-sync info which Mistika apparently didn't interpret correctly. Looking at EDL there were several lines labeled '*AUDIO.RTAS'. Once those are removed and EDL reconformed the problem goes away. This also happens in AAF, not just EDL.
  21. jan1

    Spanned Media

    I'm currently importing a number of older projects into Mistika and rendering them back out, mostly to make sure I have the workflow down before I use Mistika on my first client project on a deadline. In addition to the project above, I ran into trouble on another one. That one was filmed on a Varicam, edited in MC. Very straight forward interview footage and some titles. Created both AAF export and EDL. Both import fine into Mistika, AAF does a better job with the transitions. On the EDL things get chopped up that requires rework to avoid extra grading. But the bigger issue is audio. Both imports, when playing them have a severe audio lag. First I thought it may be a delay in my playback, but when I render the timeline out the playback persists, so it's real. I didn't find any parameters to adjust. The audio is just part of the main video files. I checked and there is no sync issue in Avid, so I'm certain the files are good. When I look in attributes I see both the proper file listed for video and audio. Maybe specific to this particular codec. I'll see if it works better with footage from another camera.
  22. jan1

    Spanned Media

    I ran into an issue today, checked the manual and didn't find the answer. I was bringing over a project via AAF from Avid. The media was XDCAM with some longer clips which were split into segments. In Avid using AMA these are automatically concatenated into a single clip. Thus the AAF referred to them with the base name (1st segment). The other segments imported into Mistika as separate .lnk files with the _2, _3, etc. After the conform everything that was past the first segment showed the offline checkerboard not surprisingly. I tried using the 'Import Sequences' in the media browsers to see if Mistika would recognized the spanned segments, but that didn't work. I guess sequences is only meant for image sequences, where it works fine. Of course the other solution is to have Avid create a transcode as part of the AAF export, which works for the video, but surprisingly loses the audio files on import. And of course also takes much longer than a linked AAF export. Any best practices? Thanks, Jan
  23. Great discussion, will have to play with the title tool more. For similar reasons, my current go-to is designing the titles in Illustrator (a bit more text layout control than Photoshop), and then drop it into a Fusion animation to convert it to a ProRes4444 file that can be imported. In Fusion it was easy to create an expression for the Y coordinate that moves in even pixels. That's always the crux of animating an externally designed titles. In Fusion you can put this expression into the Y axis of a transform node: 'time * 0.0005 - 0.5' if you're crawl was 2000 frames long as an example. And you would make your external graphic an even multiple of the frame count, either 2x, 3x, or 4x. Is that something that Mistika can do - express values as expression that reference other data? Haven't had a chance yet to dive deeper into the Comp3D tool beyond applying alpha channels. But being able to do all the animation and even some text inside Mistika speeds things up rather than switching between multiple apps.
  24. jan1

    Simple Comp

    Thank you Cristobal & Yoav for the explanations. That makes more sense now. It was indeed a bit confusing to have to put the overlay at the bottom of the stack, especially since it comes in on top during the conform. I got it to work.
  25. Thank you for the response and explanations, that makes sense. Totally understand that NAB is first priority, I was just collecting a wish list. Also, I have found some of the older longer tutorials on Vimeo which are helpful as well.
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.