Dennis Nikolaev's Activities

Slow down the automatic rotation in the Acute3D Viewer

Dennis Nikolaev — Sat, 23 Jun 2018 10:17:40 GMT

Hello, I am building a promotion display for my Photogrammetry service. 4k TV will display 4k videos captured from many of my .S3C files

I am using a screen capture over Acute3D Viewer to create videos of the projects I completed.

The only trouble is the automatic rotation speed is way too fast for my taste.

I didn't find any option to slow it down in Acute3D Viewer. So I did it in post-production on the video. However, It became quite jerky.

Please, help me to figure out how to get a SMOOTH SLOW video out of my .S3C models.

Ask A Question II

Sat, 23 Jun 2018 10:17:40 GMT

Ask 10 questions in a forum.

Irregular shape production

Dennis Nikolaev — Tue, 27 Mar 2018 11:19:20 GMT

I just say one of my competitor's CC model. It was nicely trimmed according to an irregular shape of the site.

Does anyone know how to do it?

Building a server for CC?

Dennis Nikolaev — Mon, 16 Apr 2018 10:58:52 GMT

Hi James,

I'm getting ready for a big project. And building a super PC.

The quote I get from my IT is for a

Double processor Xeon E52673 V3 @2.4Gz

6x GTX1080Ti cards

+...+...+...

The quote is about $17 000. Before I spend the money, I hope to ask

1. if CC can utilize such a monster?

2. Would Double processor of Xeon W21-45 @3.7 Gz be significantly better?

3. If CC cant use it, am I doomed to buy 6x PCs with 1 CPU+1GPU? + (CC Center licenses)

LARGE scale project

Dennis Nikolaev — Thu, 22 Mar 2018 10:43:55 GMT

We are looking to bid on a 3D model for an entire CITY!

Now have only standard edition license, will be needing many Center licenses.

I have a few questions for anyone who knows at least something on the subject :

1. What is the largest area (project) produced using CC (km2) cm/pix.

The largest I've done was about 15km2 it took forever to shoot and process. Now we are looking for 350 km2 of an urban environment.
I wonder if it is realistic at all.
On many big data sets, CC just gives me a finger. Even if AT is done properly, I get plenty of missing tiles. Smaller projects under 1km2 is a piece of cake. A project this large sounds scary. How big is the largest done in the history? If the info is available: how many pix were taken, average height? resolution?

2. What kind of image acquisition system is optimal for 350km2?

On smaller projects, just phantom 4 pro is doing just fine. Models come out a bit wavy but with models within 1 km2 it is hardly noticeable.
Acquiring accurate GPS for images is critical. Collecting control points for an entire city is not a good option.
RTK seems to be a must, with 5-way high rez camera system. Right? What kind of platform offers it? Preferably a UAV. (the tallest city structure is under 250m)

3. We'd have to build a comuter farm for a project like this. Is it possible to process the whole thing as 1 piece? What is the largest CC farm known to exist? (no need for specific names, just a rough idea)

4. Any advice on processing something this massive?

5. Is there a viewer that may allow layering information into a LOD model (like street names, building owners, zoning maps,..ets)

Ask A Question I

Sun, 27 Aug 2017 10:10:53 GMT

Ask a question in a forum.

CC Multiple GPU

Dennis Nikolaev — Sun, 27 Aug 2017 10:10:53 GMT

How many GPUs is possible to put into a PC to maximize CC productivity on Standard edition?

Ultimate hardware set

Dennis Nikolaev — Sat, 14 Jan 2017 10:51:47 GMT

Just paid for my CC license. It is REALLY expensive.

Now, I need to find an ultimate computer to extract the most value out of CC. I know there are general recommendation, like, get the best GPU... I'd like this post to be about specific specs.

Like if money didn't matter for hardware. What should I get (January 2017)?

ContextCapture AT multi-pass effect on final result

Dennis Nikolaev — Sun, 29 Jan 2017 02:15:55 GMT

AT setting for Component construction mode there is a setting for "one pass" or "multi pass".

Just now I was reprocessing some of the previous jobs that were done at "one pass" before. Now I run them with "multi pass" setting.

I noticed a reasonable improvement in details. As release 4.4 made AT stage more efficient - the increase in AT processing time from "one-pass" to "multi pass" is not prohibitive even on large sets. I'd say a few hours more, not days.

Therefore I consider it "worth it" to run all my jobs at "multi pass". Considering that a large set may be processing for a week, a few hours extra spent on "multi pass" to squeeze a few more details out of CC is definitely worth it.

The added details come from "bad" zones there overlap is not ideal. What I also learned is that there is NO IDEAL set, especially when it comes to urban environment. No mapper how well you cover the area, there always tight spaces where the camera didn't shoot enough. "Multi pass" reduces the size of those "bad" zones or sometimes eliminates them.

I'm so excited that I recommend to include the "multi pass" into default settings instead on "one pass"

One more time "huge thanks" to the team for this miracle software.

Can't wait for neural networks elements to be integrated into ContextCapture to recognize water and other common elements.

Dennis

ContextCapture CENTER Econmic reasoning

Dennis Nikolaev — Sat, 28 Jan 2017 11:33:10 GMT

I came across several bids to do LARGE scale mapping 20 - 40 km2. (Lost them, but the trend is there)

So I am preparing for scaled production in case I eventually win such a bid.

The big question is if CC Center is necessary?

CC Standard edition does up to 100 Giga pixels. That in my experience can cover 10 -15 km2.

Let say I get 40 km2 job. Then I can manually split the job into 4 photo sets (no big deal). Then run it on 4 separate computers running 4 separate CC Standard licenses. Then use control points to stitch 4 productions together.

Considering a speed of 10 Gigapixels per day, 100 Gpx finish in 10 days. 400 gpx split to 4 computers finish in 10 days. Plus AT plus a few days to shoot it. I can finish a 40 km job in 14 days - 2 weeks. That might give me what I need to win the bid.

To accomplish this I need to buy 4 CC licenses. Let say the cost of one CC license is 1X. So I need 4X dollars to make it work.

Now if I do the same with CC Center:

CC Center is 7 times the cost of the standard license (7X). Then each engine is (5X).

To accomplish same task with CC Center I need 7X+5X*3=22X (1 License for the center edition + 3 engines)

To do the same job I can spend 5.5 times more money and do it on the CCCenter.

The advantage is that I might do less stitching (maybe). I'm actually not sure is 400 Gpx can come out of AT just fine to be split later and how long will it take. So I'd have to split it in sets anyway, so AT goes smoothly.

Please, tell me that I'm wrong and it makes sense to pay that much more for CC Center.

Otherwise, if all same work can be do with the Standard edition, feel free to delete this post.

ContextCapture photo acquisition app

Dennis Nikolaev — Sun, 18 Dec 2016 12:24:12 GMT

Hello, I'm new to CC.

What photo acquisition app do you use?

Before CC I was using Pix4d (another photogrammetry soft). They have a nice "Pix4d Capture" app. Fire up the drone and let the app collect images along the grid. It still works for me in CC. However, there might be a better app.

So, Anyone, who does automatic image capture for photogrammetry, PLEASE, share, what app are you using.

Dennis

ContextCapture Paralel Viewer work around?

Dennis Nikolaev — Sun, 18 Dec 2016 12:38:25 GMT

Hello, is there a work around to display a LAYER functionality in the CC viewer?

1. I got a client who wants to see his design inserted into a 3d reconstruction. OK, I can do it in a separate file. However, I would love to turn a few original tiles off and inserted OBJ on. The on and off , on and off from different angles.

2. another problem I have with a viewer - it doesn't offer a paralel projection. Yes, I can get an Orthophoto, but it is not quick and doesnt seem to be significantly better then the model itself. The model is super fast. Besides, I cant get orthogonal building elevation with othophoto. Got to exoprt a model into a different software package, then do the paralel projection. It is a bit of a pain.

Anyone has an Idea how to do it in a better way?

Dennis