Urgent: Duplicate Your Matterport model on iPad before doing Cortex Merging11957
Pages:
1
 WGAN ForumFounder & WGAN-TV Podcast Host Atlanta, Georgia |
DanSmigrod private msg quote post Address this user | |
| @RichardStanton Thank you for the alert/recommendation ... Hi All, From this WGAN Forum discussion: ✓ New Matterport Capture Version 4.0 Issues? Quote: Originally Posted by @RichardStanton --- Here is the entire post: Quote: Originally Posted by @RichardStanton Best, Dan |
||
| Post 1 • IP flag post | ||
Hilton Head Island, SC |
THRHHI private msg quote post Address this user | |
| The two scans I had trouble with both went haywire when trying to do Cortex. For fhe last week I have been duplicating all my models before attempting Cortex. That way I have at least 1 usable copy. | ||
| Post 2 • IP flag post | ||
 WGAN FanClub Member Queensland, Australia |
Wingman private msg quote post Address this user | |
| It is not just that. On the last Friday I was in 360 mode but the capture app was doing 3D scans instead. Plus I had many situations where a 360 was captured but not even generated. I did a lot of 360->3D then and I made a copy before I started using Cortex. Not because of the described above issue but because any 360 that was placed with a bit of offset to its actual position was moved to a wrong spot after conversion by the app ruining mini map. Deleting it still left some data on the minimap. I do not think this last one though is caused by the app update, it has happened quite a lot before. |
||
| Post 3 • IP flag post | ||
|
RichardStanton private msg quote post Address this user | |
| @Wingman I can confirm the same issues that you were seeing. It's clear that there is more photogrammetry entering the mix even in the standard scans. I have started to notice more of the cortex signature radial scan line bleed into the texture mapping , even going outside of the IR outer perimeter during normal scans. It looks like one of the main collateral impact points is the local preview map. Even nixing cortex conversions all together I find I am still ending up with an extremely corrupt or "frayed at the edges" look by the end of the capture. They had a lot of us in the beta app - I am not sure why this change in particular completely skipped beta and went seemingly straight into production. I know that the cortex engine uses it's own release mechanism.. which very possibly does not have the functionality for multiple branches outside of dev... I have also noticed an updraft of random phantom clipping objects in the model and nav points "floating" far more than usual (sometimes on aforementioned phantom geometry). This bricked model thing, however, takes the cake - whatever is corrupted appears to be asset based as I am able to transfer the issue around to various hardware with only file system changes. Unfortunately I have yet to figure out where the culprit is due to the level of obfuscation they have in place on most of the binaries. Oh, and I too have seen a lot of the 3D scan to /dev/null behavior... it's unclear to me if it's actually saving something, and in hindsight I cannot recall if the next scan that does appear after a bout of that has incremented. I should know that too soon, though, since I will have a microscope on the file system for each scan anyway. I have also noted, however, that at even in the case of these bricked scans the issues discussed seem to disappear in Post for the MOST part (I still see some of the phantom geometry, floating navs for instance) I think I am going to use a secondary device tomorrow, hashing and trying to force a reproduction given I cannot even speak to support until Monday. While I do not necessarily mind a little cortex in the mix for regular scans it has been a growing concern of mine for commercial/BIM clients. We all know cortex plays it loose... VERY loose on alignments. If that same behavior is creeping into the core MP2 scan it makes me wonder what that might be doing to the point clouds. For now it's not worth the $$ for me to purchase a matter pack just to test, but I am getting very close to that point. I'll report back anything of substance or amusement that I find! |
||
| Post 4 • IP flag post | ||
|
WGAN Forum Founder & WGAN-TV Podcast Host Atlanta, Georgia |
DanSmigrod private msg quote post Address this user | |
| @THRHHI @Wingman @RichardStanton Thanks for keeping the WGAN Community posted on any challenges that you spot. Dan |
||
| Post 5 • IP flag post | ||
|
|
RichardStanton private msg quote post Address this user | |
| All, I talked with Matterport support yesterday regarding the issues above and I wanted to share the takeaways: * Mini-map going black/corrupt and no longer being able to edit the model - I confirmed that this has happened to a sizable group beyond just me. - They are working on a fix for the minimap to be released in a hotfix TBD. - There is no conclusion on what causes or exacerbates the issue at this time - I shared my thoughts regarding cortex translations and they will be getting back to me with any likely mitigation factors after talking to dev. - THERE IS NO WORKAROUND known at this time! My personal suggestion would be to make copies of models before doing any kind of edits, especially cortex driven 3d tranlations. * Mini-map clarity, positioning, scan-bleed, scan count/response etc. - I confirmed that these issues are known and being tracked in jira - They are working on a batch of updates to address these issues and more with the minimap since the 4.0 release I am still investigating mitigation and workarounds, and will report back any progress, but for now, keep your captures safe! |
||
| Post 6 • IP flag post | ||
|
filinto private msg quote post Address this user | |
| Yes the same issues with my Capture App. I want to move some scans to a different Floor (because they have been automatically misplaced) and when I open the Capture app on my iPad Pro it shows up "black" with the message on top "updating map". It is so frustrating... I wish they could make a way to perform the editing capabilities that you can have on your iPad App to a browser console (even inside my.matterport.com). |
||
| Post 7 • IP flag post | ||
|
|
RichardStanton private msg quote post Address this user | |
| Yes, welcome to the black screen of death club @filinto! They have asked me for copies of my jobs which have been impacted of which I have two solid examples. They weigh in at a hefty 7 gigabytes for the two jobs and a "duplicate" (which also fail) as reference points for their dev team. Here is to hoping that having a few full data sets from impacted captures will help them determine the cause (and work out a fix)! On the good side, upon closer inspection of the assets I see most of minimap elements intact in the file system, so I am hoping this is just a reference issue of some sort. As I noted earlier though, new scans to the space actually create new data on the map (just like a new blank "floor"  . Since the engine doesn't actually USE the floor designations that we input (they are purely for local visual purposes only, actual floor assignment is done in processing) I am hoping that some kind of workaround to re-build the minimap will be possible. I will report back anything that I hear as a result of the example data I sent them.Good luck all, and remember to duplicate and backup your captures! In the spirit of that statement a quick aside: I am sure others have been experiencing the new "issues" with the online space editor, like labels not showing to external users even when it has been enabled in the space settings as an example. I have found that every one of those that I have encountered has been resolved by switching to the iPad/mobile version of the space editor and doing my edits there. Looks like they tested the mobile experience more , which would make sense since that was the focus of the release... I have paired a keyboard and mouse to my iPad Pro just for this function and it has saved me a LOT of time and stress worrying about my production spaces not showing correctly or as defined in the editor! ^--- just a quick tip for those experiencing these issues. Cheers! |
||
| Post 8 • IP flag post | ||
|
PieroBortolot private msg quote post Address this user | |
| I downgraded from 4.0.1 to 3.0.10 (because unstable) I transferred backed up data back to my iPad (20 jobs) I see "updating map" for hours and nothing.... ...and later a black thumbnail How can I resolve it? I need to reupload some jobs Or is it a death valley without return? Thanks |
||
| Post 9 • IP flag post | ||
|
dave3d private msg quote post Address this user | |
Sadly we've lost a couple of jobs to the Matterport update gods as well. We did a 9 hour capture Mother's Day (2 people 9 hours) and uploaded I think that night. (Sunday). Monday morning wee hours they did their update. We were OK till we changed the name of the scan then "poof" gone. Took about 4 days to reach a live support person and they said yep....known issue with the update...sorry...nothing to be done RESHOOT. No fix released to date that we've been advised of. We copy every site before uploading now. I think we lost another a couple weeks before on another update. These failures weren't Cortex they were pure Pro2. Happy mother's Day  |
||
| Post 10 • IP flag post | ||
|
|
RichardStanton private msg quote post Address this user | |
| Sorry all, I was lax getting back to this thread. I just wanted to CONFIRM that the 4.0.2(44) beta build of Capture now fixes the black mini-map of death issue, including retroactively. If you have a black model, when you first load the scan you will see a message that it is processing - LET IT COMPLETE. It supposedly will keep trying until it finally does even if you interrupt it, but just let it complete. Once the process has completed you should find that you once again have access to your mini map and can continue to make your edits. I had 6 separate scans at this point which were all bricked as a result of this issue - all of them now I am happy to report are working normally again. If you have reverted your build to pre-4.0, it's up to you whether to come back or not. I will not testify to the overall stability at this point.. but I will testify to the fact that this particular issue appears to come to resolution. This same fix will likely address some of the issues with sideloads of job data, but I have not played with that as of yet. Good luck all! |
||
| Post 11 • IP flag post | ||
|
Hilton Head Island, SC |
THRHHI private msg quote post Address this user | |
| I hope that is the trend. Did a scan the other day and having learned my lesson previously I marked up the scan and then duplicated it prior to doing Cortex conversions. Went into the copy and did 3 of the conversions (yes the app crashed, reload, reconvert and it works fine). On the 4th Cortex conversion it went to the "realms of deep dark space, probably never to be seen again". No problem I thought and went to the original scan but when i brought it up it had also gone completely black. Luckily I had done the markings so I just did an upload of the black screen. Scan processed in 90 minutes and came up fine. I would have liked to have the conversions in there but it was better than having to rescan the property. Will be fine as long as I dont need to edit it again. | ||
| Post 12 • IP flag post | ||
|
|
RichardStanton private msg quote post Address this user | |
| @THRHHI If you still have the scans that blanked out on you can you report back if they correct themselves after getting the latest build from TestFlight? If you have multiple devices you can also side-load a broken job and it will come over to another device DOA - I have confirmed that moving the assets from place to place bring the problem along for the ride. If you aren't doing the beta, no worries, I am mostly curious to see if the full-asset scan that they implemented on first open now fixes it in other cases as well. If not I would still be curious to know once the release goes GM! |
||
| Post 13 • IP flag post | ||
|
|
RichardStanton private msg quote post Address this user | |
| @PieroBortolot from my experience on two iPads if you put yourself back to the beta you should be good to go as of now if you still needed those jobs back. The change that they made appears to be a bit broader than just a fix for this issue. They changed the behavior of first load to do an asset assessment/load from the ground up. This apparently solves several issues. You could always then just downgrade again if you felt it necessary. | ||
| Post 14 • IP flag post | ||
Pages:
1This topic is archived. Start new topic?