Jeff Prince

Thanks. I think the mesh leaves a lot to be desired compared to my other projects, but it was useful and fast.

Thanks. Yes, I’ve used the VW cloud services in the past. Unfortunately, it doesn’t not produce acceptable results compared to other software solutions using the same set of photos. For the residential job, I used VW Cloud and Maps Made Easy with the same photo set. VW Cloud couldn’t produce a usable result. Maps Made Easy did a good job given the quality of the photos due to time of day and wind/temperature issues. It was 105 degrees when I shot the house, which is at the operating limit of the drone. My ipad and iPhone both overheated while flying the drone and taking ground based photography. The drone itself started having temperature issues too along with the problems associated with flying in high temperature air 😞. For the Marina, I used AutoDesk Recap to create the model, which did a fantastic job on producing an accurate mesh, especially considering it was flown 90m altitude on a rainy day. Both of these examples are from worst case scenario flights. I suspect the softwares are only going to improve moving forward and hope for more automated processing that can determine the difference between architectural features and classify them accordingly. Hopefully VW Cloud will realize how useful this process is on developing projects and up their game so we can do better work faster 🙂

Yes, I used the drone generated mesh to create a 2D plan view (which it did great on) and for 3D feature levels (which on this project left a lot to be desired). On the marina project, the levels were within 100 mm as compared to a survey, so good enough for initial conceptual work seeing as the survey was not finished until after the design was done 🙂. The current level of drone photo reconstruction I am using does not built the BIM model...yet. I’m guessing that within 5 years the technology will have matured enough to do this to a certain degree. High contrast architectural features can be processed with a reasonable level of accuracy, but low contrast images in the same color families are real challenging as in this residential example. Vegetation is really problematic, especially on a day with a little breeze due to the plant movement. White architecture is really difficult if the lighting isn’t good. For the high accuracy stuff, I’ve used a Faro laser scanner. Unfortunately, that equipment is cost prohibitive to use on small projects where the level of detail is unnecessary. Digital surveys are nice, but there is equipment, data collection, processing time and costs that make it impractical on projects like residential landscapes. On the commercial job, I can sub out scanning or surveying, with a survey being necessary for the civil engineering and property legal definition. For the quick and dirty site analysis, the drone photography has proven to be the fastest and least expensive method of getting basic usable data for conceptual design. For the residential project, I had 30 minutes of drone time and 15 minutes of site photography to collect enough information to then build the existing conditions in about 2 hours. So say 4 hours total with the end results including a drone mesh, a 2D geotiff, and manual modeling of the existing conditions in VW. Had I measured everything, I probably would have been doing about 4 hours of site work given the angles of the features and such with the end result being a 2D drawing and a bunch of ground based photos.

Applying this technique to a residential remodel project. Aerial Survey -> existing 3D context -> new design. It's quick, it's dirty, but it works. First time I haven't created an as-built manually. Now if only VW cloud services could produce the model from the aerial photography instead of having to use a 3rd party... Here are a couple of animated GIFs showing the progress from: 3D aerial based mesh to 2D ortho photo to modeling existing conditions to a quick redesign of the pool deck, etc.It's amazing how fast this workflow is. Like I said, it's dirty, but it works. I took some context photos from the ground for visual measuring and correcting the predicted aerial errors on this complex site.

@creatrix I just downloaded this one as sketchup 2016 and it imports perfectly. https://3dwarehouse.sketchup.com/model/c79ed5b2-6bc7-4806-9151-87c378d30061/suburban

@malonso here's a quick visual example of what has been said. I made this a while back to communicate what class overrides do.

@barkest Nice work, great to see you helping out!

Is there a place to view the classes offered at past design summits? Reason for asking... I’m curious to see what they are like in an event setting and how in depth they go to determine if I want to go.

@martinfdc Good advice!

Nicely done! This thread has become a master class on modeling pools thanks to your video.

@Benson Shaw Thank you so much for breaking your steps down! This has helped immensely! I did another unrefined test to learn your method and made it a little more complicated with a third level rotated at 45 degrees to the main axis. Worked like a charm. I owe you many beverages of your choice! And thanks to @Mason2152 for getting this conversation started, I've learned a bunch thanks to it! Exciting times!

I've been pretty happy with my macbook pro, but I miss my 27" iMac from a gig I just left, hence my desire to get myself a desktop. I just ran the Cinebench R20 test on my macbook pro and it scored 1346. I don't know if that's good or not since the scores do not seem to correlate to the thread here on the R15 test, but I've been happy with the performance...until I started playing with TwinMotion.

Yes, and this seems to echo Jim's hardware thread. This article favors the 3.0/10 like @zoomer and supports the argument by saying the RAM could be upgraded in the future. https://9to5mac.com/2017/12/15/opinion-10-core-imac-pro-sans-upgrades-best-bang-for-the-buck/ @Hans-Olav thanks for your comment. Based on your experience and the benchmarks, it seems like a wash unless you're rendering all the time. These seem like the choices now... $3850 iMac 3.6/8, 32 GB RAM, Vega 48/8GB, 512 SSD with a predicted replacement in 3 years $6350 iMac Pro 3.0/10, 32 GB RAM, Vega 64/16GB, 1 TB SSD with a predicted RAM upgrade in 2 years and replacement in 5-6 years. I imagine the Pro would see a few more years of lifespan, but it's all a guessing game to me. I hate shopping for computers on a budget.

Excellent @Benson Shaw This is the magic I was looking for. Thanks so much for explanation!

yes, this is what I did, but had difficulty with fillet on the sloping transition.

I've been kicking around the idea of buying an iMac or iMac Pro now that the new Pro has been announced. The goal is to stay under $6000 and get the best bang for the buck. A maxed out 27" iMac is $5249 A lower spec 27" iMac Pro is $5949 (slower/higher grade processor, 2x Vram, 1/2 storage) So, what would you buy and why? If going the Pro route... Is the 3.0 GHz 10 core worth $800 more than the 3.2 GHz 8 core? Seems the money would be better spent on upgrading to the Vega 64 (+$550) and 64 GB Ram (+$400)

Very nice! How would you handle a sloping bottom depicting a transition from a shallow end to the deep end?

I decided to throw a little landscaping around the pool and test it in TwinMotion this morning. The textures came in from Vectorworks, plants and people added in TwinMotion. Crazy how fast you can do things these days compared to the 1999 video I posted earlier!

Thanks and no problem. I've been learning a lot of VW skills simply by reading this forum and trying to figure out how to do what people are asking help with... if it is something I envision needing to do in the future. This pool thread for instance has helped me a great deal in preparing for my next project rather than trying to figure it out under a deadline. Trial and error takes a lot of time, so Jonathan's task based tutorials are a great time saver. If you are just starting out with 3D modeling, quality instruction gets you going on the right foot and will avoid lots of frustration. Good luck on your journey!

So, lofting didn't work so well, probably due to the changing number of vertices in the defining curves I made. Jonathan fixed my account problem on his website, so I tried out the tutorial @fabrica mentioned this morning: https://learn.archoncad.com/blog/2011/02/28/sst_1102-creating-a-swimming-pool/ @Mason2152 A membership to Jonathan's website is a great investment. The first pool I did took a dozen iterations to get it in the ballpark using my basic understanding of VW and using the help menu. The second pool took about eight tries to get it close. This morning, adapting techniques from Archoncad, a single iteration to produce the model below. It took me an hour from scratch, including reading the tutorial, texturing with the VW included library, and rendering. Getting closer, but VW did not know how to fillet the transition between the deep and shallow ends (probably due to the edge falling on a curved wall). Note aforementioned bottom fillet fail at transition. Section is watertight and well formed Nice solid model detailing...Bull nose radius on the entire curved coping was flawless. Tile inset was done by forming a solid, duplicating it in place, subtracting one from the pool wall, and placing the remaining one on a separate class with different texture assignment.

Practice, practice, practice 🙂 Most of these 3D programs work about the same, it's just a matter of getting used to the workflow. I'm going to try a loft with horizontal cross sections and see if that does the trick. Basically, take the pool outline, copy it down in Z a few times and reshape to reflect the walls and pool bottom. Not sure if it will work... Take a look at this. Boats are complicated too.

These are the kinds of complicated shapes we are talking about modeling. Nice, fluid curves, walls that have a fillet where they meet a curved floor which slopes to a drain. I hope someone has some pointers on how to form these complicated surfaces. The skatepark tutorial video just isn't enough...

getting closer, but it is still difficult. The subdivision primitive Cube helped with creating the sloped transition between the two levels.

Anyone interested in helping out on this thread?

I think the tricky thing for me is trying to figure out how to take a constructible plan and section and making it into an accurate model. The contractor needs drawings with control points for layout. These curvilinear pools are based on center points and tangent arcs, something I really struggle with drawing in VW, but came easily in AutoCAD. I need VW practice too, hence my interest in your question. Hopefully some of the real gurus here will chime in and give us a hand because the sample I made in VW does not lend itself to creating useful drawings for construction.