Love the UI…simple yet slick and fresh. Performance for me so far is great. Looking forward to really pushing it.
Thanks Dwayne!
Please post any further thoughts on how you found the initial experience and figuring out where things are. Our expectation is that most folks who have applicable experience of say 3dsMax or Maya will find things where they expect them to be. The question we need to answer is how to best present these concepts to new users who have a different background.
-i
This is a really small nit-picky thing…admittedly it could be my own OCD. When expanding rollouts such as the Meshing, Advanced Meshing etc. the rollouts generally expand up and away from the mouse position as opposed to down from it. I find myself moving up to the things i want to interact with as opposed to continuing down the interface.
Again a very very small thing.
It is a feature 
Basically if there is not enough space to show all rollouts at once, when you expand a collapsed rollout, the whole UI scrolls up so you end up seeing what you just expanded. Otherwise the newly expanded rollout would end up hidden behind the bottom panels (Transforms etc.) and you would have to scroll down to see it.
The Point Sources area is also defined as “flexible” and it will expand and contract as needed to provide more space for other rollouts. On large monitors at high resolutions, the Object Properties panel is usually tall enough to provide space for all rollouts (and you can slide the Transform panel down somewhat). In that case, the Advanced Meshing, Zhu/Bridson etc. expanding as you would expect them - down.
Once you have selected your Point Sources, you might want to collapse the Point Sources rollout to get more space for the rest…
I’m getting some great results using the Zhu/Bridson meshing method at 1 cm radius, but it is creating a double-sized mesh which doubles the polygon count and file size. I will try the other methods and post results from the same dataset.
Here are the results from the other meshing methods. All methods generate a double-sided which I don’t care for, but the results look decent. I prefer a single-sided mesh, though.
Thanks for the screengrabs!
You are right, in that Sequoia only generates a double sided mesh. We certainly recognize that it can be a drawback, the not the least of which is the increased triangle count. As development continues on Sequoia our intention is focus on addressing these issues with a variety of new meshing techniques, mesh post-processing techniques, and modelling tools.
We have a lot of ideas, but what do you think our focus should be on?
-i
Here is my wish list to supplement and significantly improve our visual effects workflow:
- Extract watertight single-sided mesh from point clouds
- Mesh reduction with ability to keep detail on non-flat areas or interactive retopo feature to significantly reduce polygon count (similar to Mudbox Retopo, Zbrush ZRemesher)
- Ability to project textures from panos onto reduced/retopo mesh
- Automatic UV or PTEX creation of normals map, displacement map, ambient occlusion map, and/or bump maps allowing the user to work with lower poly count, but maintain detail
- This is on our roadmap. We don’t have a timeline yet, but it is on the way. We are also considering primitive modelling approaches (eg. fitting planes, cynlinders, and boxes to points). If this is interesting, then please let us know!
- Have you used the current Reduction feature? Increasing the reduction tolerance (in the meshing panel) will trigger the decimation. The higher the tolerance the more faces will be removed. The decimator does consider surface curvature when deciding to remove faces. If you have used it and found it lacking in some way… let us know what you wanted it to do!
- Projection is there, but the workflow is a bit rough right now. We are currently working an alignment feature to help align the projection and when that drops it will come with further improvements.
- PTEX creation of maps is definitely interesting and we have done some preliminary work here. We will have more on this early in 2015.
Thanks,
-i
I miss the nav cube in Frost. I’ve been trying to get my mesh to rotate with various ctrl/alt/shift and Left Button/ Right Button combos - it could just be the build since I haven’t had problems with an earlier version of Sequoia. The View icon for top, side, corner views is also nice to have.
I like all the mesh build info on the right panel and radius suggestions. The blue on blue animated stripe on the progress bar is a difficult to see - sometimes I incorrectly think meshing is done.
Please check with control settings you are using right now. Under Options > Standard Controls will give you 3dsMax like navigation controls (eg alt- mmb to orbit), or should. With the new build that may have been reset. We are working on a design for a view-cube helper and I expect you will see it in SQ before we are released. Especially for users that typically use the nav cube, this will help dramatically. Our intention is to allow a full remapping of these shortkeys, hot keys, and navigation controls. If there are control preferences that you prefer we would certainly like to add them as 3rd or 4th profile.
We can adjust the color to make it more salient. Thanks for the feedback!
-i
I know every 3d software package has basic shortcuts for navigation but Sequoia is eluding me.
Where is there a shortcut menu, or help menu showing the shortcuts? I am very used to the traditional QWERTY shortcuts in Maya or Zbrush. While I don’t expect them to be the same, I am struggling with the constant RMB to go to move, rotate, scale etc. Is there a shortcut list somewhere I am missing?
Although, I can’t stand the navigation cube approach, I would really like some simple camera navigation shortcuts. Knowing what the camera shortcuts are immediately makes navigating the camera a lot more user friendly! (such as alt + LMB for maya or alt + MMB for max for orbit)
One other thing which is kind of picky… when I mesh the point cloud, it creates a duplicate object. Not sure if this is a bad thing, but it seems like the point cloud should be hidden by default or just converted vs. keeping both.
Also agree with most of the wishlist above.
Great product so far!
EDIT: I see now it is under options -> keyboard bindings. One suggestion is to make the keyboard shortcuts appear upon mouse hover. How do I orbit the camera with a shortcut?
Thanks for your feedback!
Our of curiosity, have you downloaded the Sequoia Reference PDF from the Builds forum?
I have. Making my way through it.
Shortcuts which appear when you hover over icons might help, but it looks like some of the shortcuts are listed in the menus next to the commands, which is good.
Interesting you guys based your software on the Max template of commands. Did I see there was a way to switch to Maya based shortcuts? Will investigate this further.
Thank you for the prompt reply.
Nevermind about the Maya shortcuts: Found it.
Since you can make your own shortcuts, feel free to remap everything to your liking. 
Shouldn’t take more than 10 minutes.
You could even share a saved keyboard mapping file with us if you want to consider including it with the release version…
I would love to see some pointcloud filtering like delete outliers, delete disconected and a uniform sampling filter to get a consistent pointcloud density.
I agree about the stray/noise filtering and uniform filter, because not all registration software is capable of providing such filters for every export, although I wish the opposite was true.
Took me three iterations to get these results, but it was one of my very first try of the new beta:
Argh, hard to do this justice with these re-sized low resolution images. This was a very noisy data-set that is not perfectly registered. Still the best mesh I have ever created from this type and size of data 
Greetings all,
This looks like the place for first impressions so here are mine after the first few days. Don’t read to much in to the order of observations. It’s mostly just the order I encountered various things. I have a wish list that’s a mile long too but I’ve tried to limit my comments here to stuff that is in the current build, or at least hinted at. So without further ado.
No linux build? Ok not really an issue since most of our workstations are stuck on Windows due to other software. I would love to be able to migrate some of them to linux at some point down the road though so if you ever need testers I’m available.
Overall the controls and keyboard shortcuts seem basic but sensible. The fact that most things have shortcuts and I can change them is a big plus. I did immediately change to the Alternate control scheme since I prefer the consistency of always having the navigation modifier key be the same. Given the choice I would be even happier with an alternate Alternate, Houdini style, control scheme where space is the navigation key rather than alt. I know alt probably has a stronger precedence in other software, but I find it much more comfortable to use the space bar while keeping my fingers on the home row. Additional modifiers for doing things like rotating around the camera’s z-axis and box zoom would also be useful.
Manually setting the camera focus point makes for rather slow tedious navigation around big irregularly shaped scenes. I would much rather have the focus point automatically set to the point under my mouse when I initiate a tumble. Doing that makes it extremely fast to accurately navigate around any size scene. I’m not a huge fan of using the center of the view port as the focus point, but that’s another alternative.
The initial panel layout also seems reasonable, although I’m not sure Bookmarks deserve to fill the entire left column. I’ll be curious to see how much I end up using them in Sequoia since I don’t use them very often in other programs. For now I’ve stacked the Document Explorer on top of Bookmarks and given it about 2/3 of the left column. Like keyboard shortcuts I’m sure I’ll play with this more later.
I like the choice of sqlite for the project file, much better than some opaque binary. It would be great to have some documentation of the schema as a stop gap until an SDK and/or internal scripting is available. I don’t have a specific use case in mind at the moment but it is always useful to be able to pull bits of object metadata out of a project.
I really miss having a solid color ‘shaded’ mode for point clouds and meshes. Probably 90% of the time I spend looking at point clouds it is with color/intensity turned off and with all clouds assigned either a gray color or different colors per cloud, depending on what I’m doing. Shaded mode provides the same utility it does when working with meshes. It allows you to focus on the geometry without being distracted by any texture info. All manner of sins can hide under a decent texture. Turning off point cloud lighting gets you the solid color aspect but not the shading so doesn’t seem that useful.
I saw it mentioned elsewhere that there is a desire to be able to color data with arbitrary attributes. I think this would be extremely useful and I can imagine having a material system where we define mappings between various attributes and their viewport representation. A few good defaults would probably be enough for most uses but having the ability to define custom materials could be very powerful. At least initially, requiring the user to do a little programming to define custom materials wouldn’t be to big of an ask.
Speaking of materials and color, is it worth thinking about color management? OpenColorIO? Most of us who do scanning haven’t historically thought much about color management and just accepted whatever sRGBish photos we got out of the scanner. However, now that HDR imaging is more commonly being paired with scanning by all types of users it might be time to start thinking about color management more seriously. I know of at least one VFX studio that was experimenting with HDR point clouds as an alternative to image based lighting but put it on hold because of software limitations. I guess for lighting though you don’t really need color management, just a scene linear data source.
It’s to easy to move objects around. Generally, by the time I would be using Sequoia I either wouldn’t want to move scans around at all or I would want to do it in a very controlled way. The ability to lock object transforms would be an easy way to address this.
It seems like it would be useful to have a generic Group object for applying things like transforms or material properties to multiple children. The ROI object is sort of a special case of this, but in many cases the ability to clip the children wouldn’t be necessary/desired.
A few times while working with the software and reading through the manual I found myself wishing the underlying node graph was exposed to the user. For example, the way the Simplify operator is presented as part of the meshing panel I initially didn’t realize that it could be used multiple times without regenerating the source mesh. The relationship between those operators would have been instantly clear if I could have seen the node graph. Taking that a bit farther, when we deliver meshes we usually provide multiple versions, typically 25%, 10%, and 1% of the face count of the source mesh. It would be really useful to setup a node graph with a single source mesh feeding the required simplification nodes, each feeding their own export node. Of course there are other, potentially simpler, ways this could be exposed to the user. Personally, however, I would rather see development resources spent on building powerful tools rather than trying to hide complexity from us users. I also think that the way some of these relationships are currently represented (or not) in the Document Explorer is more confusing than just showing the node graph.
The list of supported point cloud formats for import is good. Someday I hope to be able standardize all our transfers on e57, but for now ptx and the other text formats are necessary evils. Ultimately I would also like to be able roundtrip data through Sequoia e57->sprt->e57 without data loss
The selection of mesh formats is also fine. Alembic is the only other one I use that isn’t on the list. I’m sure someone will also ask for fbx eventually.
Looking forward, I’ve been thinking a little about what would be necessary for me to go beyond just playing with toy datasets to start using Sequoia, at least in a limited way, on some of our production tasks. Right now Sequoia’s main function is point cloud meshing, however, from what I’ve seen so far I think I can produce meshes that better suit our needs using our existing tools (Screened Poisson Surface Reconstruction). Therefore, I’m more interested in the development of either mesh post-processing tools I can use on imported meshes or point cloud processing and segmentation tools that I can use before meshing. Of those two, point cloud tools would be higher priority for me since fewer alternatives exist in other software.
All in all I really like what I’ve seen so far and hope I can provide some helpful feedback as things progress.
Hi Jed, thanks for the feedback!
Not yet, but it is planned eventually. It appears that the majority of potential customers are on Windows or OSX, so we are trying to cater to that group first.
We might make the “Alternative” controls the default ones, but we would have to change the name I like that ALTernative uses the ALT key, and would hate to lose that implied connection. But it is worth thinking about.
I guess we could allow any key to be used in place of ALT for viewport navigation, as well as allow customization of which mouse keys are used for which operation. it would be more universal than just adding a 3rd preset configuration.
Good suggestion, will log it as a Wish. Having a set focal point can be beneficial for new users who navigate with the icons. Using Z to Zoom Extents sets the point at the object’s center, and Pan moves the point. I think we should probably display it explicitly in the viewport so it is more obvious. But having the ability to use the mouse pointer as the center of interest certainly sounds like a good idea.
Hurrah for customizable layouts!
This is already in the next Beta build (coming soon). You will have the choice of using the incoming color as before, or switch to flat or shaded mode using a custom color which can be defined per object. We also provide a drop-down list to select the Color channel source from alternative channels available in the stream. For example, setting the display to Normal instead of Color will shade the mesh according to the normal vector direction. The only current drawback is that this requires the Update button to be pressed as the data has to be resent to the viewport sub-system.
We have some experience in that area from our Draft command-line compositor that ships with our Deadline render manager. I will log this as a Wish List item.
I would counter that if you are not moving anything around, you should not be switching to Move mode at all, but I agree that a Lock Transforms option would be nice to have, as well as a Freeze option that disables hit testing in the viewport (but you could still select via the Document Explorer).
…And general parenting, as part of this.
Exposing the node graph wasn’t something we wanted to do in a v1.0 system, but it is in the long term plans.
For producing multiple versions, wedges, mass-converting files, mass-importing files, mass-exporting data and many other typical tasks that would usually require a node graph or a script editor we are working on a simpler spreadsheet-based interface that would be more familiar to the average user with no high-end 3D application knowledge. Stay tuned!
We are working on a new SPRT implementation that will get you there.
We might do Alembic, not sure how we feel about FBX. I guess like everybody else 
Once again, than you very much for the feedback, a lot of good points there!
I am off to our Wish List to process them
Given how varied the control schemes in other 3D app are this would be the ideal solution.
It can also be useful when you have a sparse point cloud and it may be difficult to get the ray cast from the cursor to actually hit a point. In that case the fall back focal point might be very far from the region you’re looking at and all of a sudden a tiny mouse movement causes a gigantic rotation. This can be particularly confusing for new users who haven’t learned the mechanic yet. Once learned though it is a very efficient way to navigate. The software I use this in most includes camera transforms in the undo stack so that also helps mitigate the consequences of an unexpected camera movement.
What about having two modes? The default would be the same as now with the addition of a hot key that enables focus under mouse while depressed. The second would be the inverse, with focus under mouse by default and fixed focus while the hotkey is depressed. I would only want to see the focal point in the viewport while fixed focus was active.
Amen!
Excellent, I can’t wait to try it out.
I won’t get on my soapbox about underestimating users. I just think that, in many cases, there is great value in showing users exactly what is happening to their data without to many simplifying abstractions. The learning curve may be steeper initially, but for any frequently used tool it inevitably pays off with greater efficiency and flexibility later.
That being said power vs. ease of use is always a hard balance to find and there certainly isn’t just one right solution so I’ll look forward to seeing what you all come up with.