Forum

Author Topic: Agisoft Vs Competitors  (Read 73276 times)

Artisan S

  • Newbie
  • *
  • Posts: 31
    • View Profile
Re: Agisoft Vs Competitors
« Reply #30 on: December 14, 2015, 05:49:10 PM »
RealityCapture seams to be a good bet for a competitor.....it has the ability the combine Lidar and PG data and that gives it an edge I guess. Personally I think Photogrammetry is great but is has a few big drawbacks:

1) It consumes eons of processor time (980ti equiped 6 core and 32 Gb machine). Now you can say buy a bigger machine but that helps only up to point as research from Puget shows.

2) The whole process has a hit and mis character that I dislike, I own a laser and a SL scanner and those systems have far more user interaction and thus far more predictability. I've seen 700 photo's scans fail to deliver a full pointcloud of a subject were an 86 camera attempt succeeded. (and vice versa to make matters worse).

3) Even working in Agisoft self is trial and error. How far am I going to drive my sparse cloud, how many points am I going to allow and how many points the system is going to deliver. How dense will my dense cloud be.....high, ultra-high of medium (like eating stake) so wait for an eon and a half with 500 photo's or go for medium and be done today, but was that an optimal result? So do it again. Or invest in a 10.000 euro workstation and open up the crap via a highest, ultra-high, maxed out, pathway. Is that the way to go?

In one word this software is driving me crazy. On one hand the results can be stuning, on the other hand it takes up so much time and effort that I wonder if it is worth it. For now I'm done with it.

Greets, Ed.

Wishgranter

  • Hero Member
  • *****
  • Posts: 1202
    • View Profile
    • Museum of Historic Buildings
Re: Agisoft Vs Competitors
« Reply #31 on: December 14, 2015, 07:06:53 PM »
Hi Artisan S

yes have right on few points but can say its about EXPERIENCE  and knowledge of the proper workflows

some videos from  very nice results where can see clearly the workflows are the key for great results...

and as we are on the Russia sw webpage, will post something that is related to Russia :D
its a RAW results no post and still not final product of scanning

https://www.youtube.com/watch?v=OgWxgZZ7pZo

 
----------------
www.mhb.sk

stihl

  • Sr. Member
  • ****
  • Posts: 410
    • View Profile
Re: Agisoft Vs Competitors
« Reply #32 on: December 14, 2015, 07:25:12 PM »
Well Ed,

I've been working very intensely with Photoscan for the past two years now and I can say to all of your mentioned points that it comes down to experience from trial and error tests. Eliminating factors that makes the alignment go bad.
As far as how many points a dense cloud gives is very easily to determine as it all comes down to the project GSD.

I've tested *all* other competitors to Photoscan and thus so far, especially the V1.2 Beta, it beats all of them. Perhaps not in sheer speed, but it's by far the most versatile software.

Wishgranter; that's insane. Very impressive. Your work?
« Last Edit: December 14, 2015, 09:15:31 PM by stihl »

Wishgranter

  • Hero Member
  • *****
  • Posts: 1202
    • View Profile
    • Museum of Historic Buildings
Re: Agisoft Vs Competitors
« Reply #33 on: December 14, 2015, 09:47:28 PM »
Hi Stihl no, not this one

 its my friend work for  State level heritage project 1+ mil scanned object from small stuff up to cities like this one https://www.youtube.com/watch?v=dmfjTSdQUG4 

have done some other work but stil have not permissions  to post it.. 
stay tuned for my new website :D
----------------
www.mhb.sk

gatsri

  • Jr. Member
  • **
  • Posts: 77
    • View Profile
Re: Agisoft Vs Competitors
« Reply #34 on: December 14, 2015, 10:10:46 PM »
Wow, which platform you use to make such amazing images?! what camera is it??
it's amazing, I love it!!

Wishgranter

  • Hero Member
  • *****
  • Posts: 1202
    • View Profile
    • Museum of Historic Buildings
Re: Agisoft Vs Competitors
« Reply #35 on: December 14, 2015, 11:52:30 PM »
PM me  on muzeumhb@gmail.com want not discuss stuff here on AGI  site...
----------------
www.mhb.sk

Artisan S

  • Newbie
  • *
  • Posts: 31
    • View Profile
Re: Agisoft Vs Competitors
« Reply #36 on: December 15, 2015, 04:01:57 PM »
Well gentleman,

So it's back to the drawingboard. Shooting, testing, shooting, testing......untill I'm reasobly shure I know what this program needs, want's, and what it's limits are (not so many I guess) judging from these amazing results. I whiched I'd seen those earlier then I would have put a dirty sock in my big mouth :-).

What this software needs is documentation (and lots of it, and by people who know how).

Greets, Ed (who was baffled by a missing staircase in scan one, and a beautifully rendered one in scan two....only difference, less camera's in the latter.....I'll be damned :-).

P.S. Yeps I found out about automatic photo quality judging the other day........it ROCKS!

Greets, Ed.

Godfery267

  • Newbie
  • *
  • Posts: 1
    • View Profile
Re: Agisoft Vs Competitors
« Reply #37 on: December 16, 2015, 04:52:54 AM »
Just to make the Agisoft team aware of it ? :-\

Artisan S

  • Newbie
  • *
  • Posts: 31
    • View Profile
Re: Agisoft Vs Competitors
« Reply #38 on: December 17, 2015, 10:56:41 PM »
Jeps,

As an ex-software test manager (everything from technical testing till usability testing) I damned well know that great software that is complicated to master, stands or falls with great documentation. Now you can leave that to the community (like open source software mostly does) or you can take that resposibility yourself (like most COTS software does), but now Agisoft (which is a great program) is hampered by lots of little SUE's (Stupid User Errors).

I know some skulls in St Peterburg must be smoking due to the pace of the developments (bug fixes are usually also packed with some added functionality).

In the basics Agisoft is deceptively simpel (great workflow based design for the major steps), but the devil hided in the details. For instance:

1) Alligning chunck didn't work although both chunks contained an overlap of 2 to 3 pictures.....why not?

2) Then I decided to give allignment a helping hand and I masked out 490 photo's in the chunck that I mentioned before didn't allign as photo's and not as a chunk.....now they alligned with ease, why?

3) Now I've used Gradual Selection and Reconstruction Uncertainty and Projection Accuracy.....I feel like Captain America talking to Stark in the first Avangers film "Speak English please....!" or Russian if that helps (I've got some Russian freinds who could translate :-)). Now the manual does give some explantion, I quote:

"High  reprojection  error  usually  indicates  poor  localization  accuracy  of  the  corresponding  point
projections at the point matching step. It is also typical for false matches. Removing such points can
improve accuracy of the subsequent optimization step."

A) What is high........as a Dutchman I think everything above 50 meters is high.....Nepalleese folks might disagree!
B) What are subsequent optimization steps.....I've got an allignment, how do I optimize, is maken a dense point considered an optimization step? Or is fidling with Projection Uncertainty a subsequent optimization step.....think Captain America.

4) Now I notice a gradual selection in the mesh view as well, that looks like Meshlab talk, I understand......pwjiew.

5) BTW, I don't need to be in Meshview to select Tools -> Decimate, Smooth and Close holes (all great for 3D printing and gamedesign), why not?

In short I've noticed that

a) You should optimize your shooting, fair enough......

b) You should optimize your workflow......also fair enough, but to what extend do I optimize my workflow both in timeconsuption.....(running everything high, higher, highest, clogs up my computer and infers with throughput and deadlines), so were can I cut corners and were should I point my resources?

For me, it's a game of hit and miss or trial and error, through a rabbit hole darkly.......and even the internet is no good as there seam to be as many opinions as users. Because youre great mesh can be my cluncker, but also because everyone has developed their own workflow.

BTW, I've tried other alternatives (Memento is easy yo use and the result reflects that, it good but not as good as Agisoft on a good day) and I'm sticking with the Russians (they are great mathematicians and fabulous programmers, I know I've worked with them). But the documentation......hmmmm, Achillesheal......

Greats, Ed.

Reno2015

  • Newbie
  • *
  • Posts: 6
    • View Profile
Re: Agisoft Vs Competitors
« Reply #39 on: December 19, 2015, 05:01:15 PM »
how about agisoft photoscan pro VS APS mencisoftware mate?
just trial last week, i can say that APS has more better orthophoto result (true orthophoto) no wrap
and it has DSM to DTM conversion, in this case i dont know how far the accuracy for DM generation

Thanks.