Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.

Topics - Steve003

Pages: 1 2 [3] 4 5
« on: October 20, 2021, 10:58:50 PM »
Its ok creating a mesh with texture but what online viewer is there for someone to view the result you send them ?
How should it be sent ?

so far I have spent 3 days trying first obj which has a separate mtl and jpg file, and such need to exist together on a pc in a folder for the jpg to work, so online upload a file is no use, unless someone knows better.

fbx embeds the jpg but I cannot find a site that will open and play it, those found I saw a black silhouette and loads of sliders that was too complex, a client isnt going to fiddle about. Autodesk viewer was far too dark , at least it showed the image, it had no ortho view as such. 

Aspose is fbx yet it made a complete disaster of metashapes fbx export, the texture was not allocated to the parts but just dumped over it.

I am beginning to think metashape fbx export is faulty as nothing is working right yet. either far too dark, or no image or mapped wrong.

I tried fbx from photoscan and opened into rhino and it was just white object.

I also need measuring capability from it. So I need to add scale, so use meshlab to get a dimension into it.

Unless a client can see the result whats the point of creating a 3D model for them ?

I seem to have discovered the hole in the entire system , no one apart from me can view the results !


searched the internet for this, then the manual, nothing on where the texture that is generated is stored.

Just what is the address to this, both for metashape and Photoscan ?   I want to compare it to the one made in metashape, as its looking good without blurs, whilst metashape same object is full of blurs and a mess.

Photoscan wins again.


Hi, an entire day spent trying to get metashape model into a viewer so someone can view it, and get dimensions from it.

obj, fbx dae all fail. fbx metashape makes seems to be a newer version than the prog opening it, as it has errors. there is no option to export as an older version .

converting to an older version is not easy, dodgy sites I avoid, autodesk used to do it.

We should bne able to just export as fbx, choose a version suited to the client, and email it them.

Not a day trying to get a reader to show it.

Just what file type are there that contain a texture or better still the 'shaded model' images seeing that textures are blurred, again await agisoft to say why blurred.

I am told obj no good as an online viewer does not have ability to place the obj mtl and jpg into a source folder. Autodesk asks for all files then shows grey.

This has to be easier than this.

I then see this thread,

 is this problem still to be solved ?


General / Give a scale to a model in metashape ?
« on: October 19, 2021, 11:37:17 PM »
I export my texture model as an obj, I get three files obj mtl and jpg, I browse to the obj and open into Rhino3D, it doesnt apply the texture.
I open the obj into online Autodesk viewer, it doesnt see the jpg.
I export this time as fbx tick texture twice , export texture and embed texture options, easy to forget the second assuming the first is the texture dealt with !   

I open this in Autodesk and I see texture, though image is dark. There has to be better than this surely ?

It has a measuring tool, it says no dimension, it hasnt a scale.  I see no way of selecting two parts in model in metashape and entering a distance.

How can I give scale to this so the measure tool works in Autodesk ?

Does anyone know of a 3D viewer that has proper ortho views, can measure and opens a fbx bright and clear.

After all my hard work I cant get anyone to view the end result from metashape and in ortho as well as perspective and measure on it as I need them to.

I decide to open the fbx into Rhino where I can scale it 1:1, it errors on the metashape fbx !

why does nothing work, all avenues barred.
Unable to scale it anywhere for that dark dingy Autodesk 'non ortho' ortho viewer.
What avenue could I try that wins ?

having run command make texture, although a texture with blurred areas is the result, when importing this obj into photshop it comes in as shades of grey and not the textured version.

I then try and again its a very pale grey with  brown areas. no texture at all, and when browsing I browse to the folder that has the obj an mtl and a jpg file in it formed from the export model command.

I show the decimated version as this viewer would not accept >35Mb, (the obj was 101Mb) I chose decimate model and got offered 2,617,884  when it said source face count 983,413 which in my maths is not 1/10th of source so I typed in 98,341 and now its very poor definition but now 9Mb. That aside there is no texture image.

What did I do wrong there by the way as the result is very crude now. and 9Mb versus 101Mb, all thanks to this prog demanding less than 35Mb.

I import the non decimated models obj into Rhino3D and it has the items colours BUT its showing the model shaded mode and not the textured, (but blurred bits) version, so again it hasnt latched onto the texture jpg.

I then try Autodesk viewer and again it opens in greys despite the promising fact that it showed all 3 files !
I decided to export the  'shaded model' type coloured model as seen in metashape, from Rhino ticking add textures though it hadnt brought in metashape textures)  and it (Autodesk) even asks if there are more files to accompany the .obj

Even that has opened as greys, so the failure of Metashape to export .obj with a texture is then causing further problems throughout the process thereafter.

Surely that jpg needs to be embedded and fused into the obj, sitting outside of it in a folder seems to be the failure in all these !

do all progs generate .obj and make the jpg texture image sit outside of the .obj ? seems a recipe for disaster.

Where are these texture images going when opening the metashape created obj  into viewers and other CAD progs ?

and what did I do wrong when decimating the faces count using my own maths ?


Metashape 1.7.4
File export model as obj name it, then up pops a panel with all sorts of settings, one is Precision and it shows 6 but no indication of what range there is and what the numbers mean,
There is nothing on this explained in the pdf manual either.
 is 1 bad, or good, is it about decimating the mesh again or what ? Does it affect the quality of the image ?
At the moment shaded view is giving a more uniform image than textured which sees some areas go very blurred.
One can wreck all the work so far unless the settings in this box are understood and explained.


Hi, I have a model of the inside area of a trolley and another of the outside area of the same part of the trolley, they will see some common features. the textured floor being one of them. Another model will be a hook at the rear, again same area.
If I had modelled each part in the same project as chunk 1,2 and 3 then I could go combine chunks.
Am I right in thinking if each chunk has already been modelled using depth maps and has texture, the combine chunks will bring the models together, or will it require mesh making all over again ?

However as such they are two projects (separate .psz files)  and I cant see a way of importing the model then combining it with the existing one.

I import the model but no chunks come in.

How do I combine the models ?


Hi, I have align photo's my agricultural tractor.
hit numpad 5 for ortho, then numpad 7 for top, rotate model until its aligned to the grid and I am looking down onto its top.
hit numpad 1 for front and rotate model (though the ball has its red/green/blue lines not aligned to ortho (how do we reset that ball ?) until the tractors front grille with headlights is facing me on screen.
Hit ctrl numpad3  (it can be normal numbers on keybd also) for Left view of tractor and I am seeing its right hand side. Is there a way of making Left be right side of a vehicle ?
I see Rhino3D also does this and that its an international agreed thing but I wish whoever invented this had it more logical.
I rotate it to see its left and now 1 for front shows its rear end. Just get used to 1 is rear and left is right etc.

I also wish there was a grid other than just in 7, for aligning to.


My project is an agricultural tractor with a glazed cab (cab has front and sides only) circa 1970's.
Paintwork is aged and not glossy.

Shots of glazed cab interior all taken from outside cab, impossible unless fisheye lens to do sat inside interior. Doors were shut. camera access from rear to some extent.

Cab interior has in photos taken through side and front of cab , parts of it spoilt with reflections of items elsewhere in the yard, HOWEVER there are areas perfectly okay where its as if the glazing was not present, these would surely be useful if they also occur in a few more shots. The exterior having the 60% overlap to tie the photos together and these also would have such.

So could the spoilt parts be masked off in the photos, each photo would have to be masked by hand ?

There are shots where the entire glazing is with reflections or just featureless and these must be masked off. Mask from model I foresee not being of use as the model with all this glass wont even get made to be sourced from !  I have asked of this catch22 a few times without answer. If such happens is it the awful task of hand masking 102 images ? That gets the model made for sure, and it would have no glazing just framework as a result.
However that would also nuke the interior, the parts without the reflections. Even if portions get modelled as I would expect, I can use those to remodel using curves and surfaces the interior, as is the intention anyway for the entire tractor, in my CAD prog Rhino3D.

How is this situation dealt with ?

Also if some photos have those coloured spots from sun getting in on the act , despite a lens hood,   and we can see the structure/surface through the translucent ones, can the affected areas/spots and ellipses, be masked out and that photo still be ok ? Is Agisoft going to fail when seeing those if they occur on the object in the photo ?

What if you almost finish the photos and it starts raining and the item has some rain spots on it, can those be masked out on the photos to enable the set to be used. I hear you say redo when it stops raining, well what if it carried on and on and going back again is not possible. ?

REAL LIFE PROBLEM. no we cant go throwing talcum powder over someones exhibit.  Often owners and museums have restored their vehicles, tractors, and other industrial equipment and gloss paint is usually used. In dull museums I have to use flash, sometimes it is the only means of even getting a photo, I pull the white sheet out from the flash head and let the light bounce off that as an aid to using existing light which on its own is too dull, and 800asa upwards adds noise which is also to be avoided ! Low ceilings if white are useful to bounce the main flash off. However the gloss , on what was a rusty item and now painted, sparkles a bit on the uneven surface.

Also a glossy old tractor wheel at an out of doors sun lit day, same areas of sparkle, Photoshop raw mode highlight slider set to 0 can help, but doesnt totally remove it, the details are burnt out at times so not there to bring back.

Can areas afflicted with such be masked out in each photo ?

I am not seeing anyone with 'real world' problems as these showing how to deal with them.


Is all of this correct ? What order should these stages and decision points be in.
It perhaps needs one of those decision and action diagrams with triangles and squares.
Anyone care to do one ?
It can be bewildering to know what mask methods if any are required and when, despite the basic workpath steps Metashape offers. masks are not mentioned in those listed steps. Masks are a users decision, and making one and saving on many hours of masking each photo is vital so knowing when and which one is what this is all about.

Workpath journey and decisions at crossroads.

Scenario, object is out of doors and immovable. Weather cloudy bright no sun or rain.

A. Assemble the required photos with shadow detail visible, no sun, no shade, no rain spots, no varying background (if background has changes use masks, at start of all this, see F.)
B. set of photos, align photos
C. perhaps orient and set up region box on sparse point cloud, using ortho views ( 5 toggles ortho, then with ortho on use 7 1 and 3 numpad keys), or do so in dense point cloud result.
D. show cameras, see there are some overlaps. ignore them, (as unable to select them) create a model then select the overlappers, right-click in the Model view and choose Filter Photos by Cameras option in order to leave only selected cameras in Photos pane.
E. decide if masking is required, choice of
E1. create mask with masking tools for one image or use e.g. photoshop and save as .tif with alpha. Apply that mask to all cameras. Metashape will mask up all images with it.
E2. mask from model, having created the model then use this to make the model. (unless model needs a mask to even manage to make it, if so use F)
F. make the Mesh, options
F1.mesh from dense point cloud, so make a dense point cloud first and delete unwanted areas on it then make the mesh from it.
F2. mesh from depth map (better definition on thin pipes, rope etc, ) no need for dense point cloud stage.  May need masks to avoid unwanted areas, see E1 or E2.
F3. mesh from sparse point cloud, too crude perhaps to make a mask selection from but allows camera positions to show for editing out unwanted cameras.
G. apply texture


 I foresee a catch22 situation.
I love the idea of saving days of masking on a 500 photo project by having agisoft be told what is the model and exclude everything else.
What if its confused with the background featuring humans, or sunlit areas with sun going in and out, or sunlit areas moving (as the sun does) during the shoot., or foliage moving etc. real world situations in fact.
Its unable to make the model which could then have all the unwanted bits removed indicating to metashape what is the model and what to mask.

Can this be so ? a catch 22 situation.

How is this situation dealt with ?

Can a mask be made in photoshop or with the mask tools and that get used to make masks for all images ? Masking every image by hand is impractical if its a complex shape such as a farm plough in a sunlit junk yard.


General / What needs doing to be able to use Reduce Overlap command ?
« on: October 15, 2021, 12:01:22 AM »
Metashape 1.7.4
Tools>Reduce overlap command.
It is greyed out, what do I have to do to be able to pick it ?
I have just aligned 155 images and threw some extra ones in to try and plug some gaps, so some are needing removal as they share same location or thereabouts.
I have obviously got turned on show camera locations.

I tried selecting all but still that command is not available.  :'(


photos of a tractor wheel, other tractors, grass, parts of its own tractor, all unwanted in the photos, align photos to see if I need to mask it all out, and impressively and  jolly good I have a point cloud, I see the wheel, along with points scattered around it of those other items, select (pink) and delete the points that are nothing to do with the wheel, then create dense cloud, I see more areas now including grass at base of wheel, so select and delete that, then generate model, and that grass is back again. yet view dense cloud and its gone, why and how can it be there in model ?


metashape 1.7.4
I am totally perplexed with the 3D viewport. I have a wheel, I rotate it until the grass its on is at its base and the wheel is upright. I then need the rotate ball gizmo red green blue lines to be reset to suit the wheel, as it is the red line is angled as are the other two and using it the wheel goes over on an angle, one cannot use the coloured lines to rotate the object as one would in a normal 3D ortho viewport set, If I want to rotate around the wheel as one might walk around it on the grass from front to back admiring it, instead it goes off on a silly angle as if there is no gravity. I need the grid to be the grassed area. I need a top view , a front view and a side view, then I can rotate the region and bring it in to hug the wheel. One needs an ortho view in three directions to best rotate and rescale the region.
I could do this in Rhino3D in seconds. There is only ortho view, but which ortho, is that top or front or side ? I want my vehicle wheel to be front for its face with hub, side showing tyre view fwd, top for a view down on its top.
The whole thing is highly unsettling, almost brings me to tears !   :'(

This is a 3D prog yet its basic 3D navigation is a killer, or is there a way, as I wish it was obvious.
Rhino3D has 4 viewports top, left, front, perspective. The left can be made right, the top bottom, the front back. double click one of the names and it takes up the space all 4 did, double click the name again and we are back to all four.
This is what we need here.

Help !


requirements for metashape 1.7.4 just states hardware, I have googled this and found nothing, there has to be a software spec shown as well.  but one also must know what OS is required, installing onto an OS that isnt declared compatible can cause mayhem.
I am win 7 64 bit pro. what is the last version of metashape that is known to work on such ?


Pages: 1 2 [3] 4 5