Open daviddrezner opened 6 years ago
daviddrezner
commented
6 years ago following is a dump of active nodes of the middle slice reshape to square image , I've expected to get corners in a circle shape locations (this due to sphere ptcloud input), and got following:
I assume that my simple reshape is wrong?
mkazhdan
commented
6 years ago I’m not 100% sure I follow the question, but the nodes are ordered first by z-value (from slice to slice) and then by z-curve order (within a slice).
-- Misha
From: daviddrezner notifications@github.com Sent: Monday, April 2, 2018 9:53 AM To: mkazhdan/PoissonRecon PoissonRecon@noreply.github.com Cc: Subscribed subscribed@noreply.github.com Subject: [mkazhdan/PoissonRecon] poissonRecon nodes order on given slice (#51)
Hi Michael,
Can you please explain the order of the nodes for a given slice before MC algorithm. given to the poisson sphere ptCloud, I'm dumping nodes of the middle slice just before the MC (Marching Squares), having problem to understand the spatial order of sValues.mcIndices[i - sValues.sliceData.nodeOffset] hence corners indicator data.
Thanks, david
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/mkazhdan/PoissonRecon/issues/51 , or mute the thread https://github.com/notifications/unsubscribe-auth/AKXc5RJqTmws90qzrxj8_IITKJsUpjIkks5tki04gaJpZM4TDj9u .
daviddrezner
commented
6 years ago Thanks Misha I will be more than grateful to get your assistance with following,
My task is to overwrite the indicator function just following the poisson solver, with outband values (negtive values), so the final mesh will not pass through this voxels.
I've succeeded to overwrite some entire indicator function slice with a negative values (outband) and to split a mesh sphere. but I've having trouble to do it for only part of the nodes on this slice, there are following error: "[ERROR] Failed to close loop" _setIsoSurface() function.
drill down: my indicator function overwrite done in the _setSliceIsoCorners() function, set squareValues to negtive value just before MarchingSquares::GetIndex(squareValues, isoValue); function this action produced the error that I've mentioned.
Is there a way in the released code to mask specific voxels to prevent mesh passing through?
That's the source of my previous question regarding the order of the nodes within the slice, is the z-curve order is only of the active nodes? what is the relation to the voxel absolute location of that slice.
Best Regards, David
On Tue, Apr 3, 2018 at 5:18 AM, mkazhdan notifications@github.com wrote:
I’m not 100% sure I follow the question, but the nodes are ordered first by z-value (from slice to slice) and then by z-curve order (within a slice).
-- Misha
From: daviddrezner notifications@github.com Sent: Monday, April 2, 2018 9:53 AM To: mkazhdan/PoissonRecon PoissonRecon@noreply.github.com Cc: Subscribed subscribed@noreply.github.com Subject: [mkazhdan/PoissonRecon] poissonRecon nodes order on given slice (#51)
Hi Michael,
Can you please explain the order of the nodes for a given slice before MC algorithm. given to the poisson sphere ptCloud, I'm dumping nodes of the middle slice just before the MC (Marching Squares), having problem to understand the spatial order of sValues.mcIndices[i - sValues.sliceData.nodeOffset] hence corners indicator data.
Thanks, david
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub < https://github.com/mkazhdan/PoissonRecon/issues/51> , or mute the thread < https://github.com/notifications/unsubscribe-auth/AKXc5RJqTmws90qzrxj8_ IITKJsUpjIkks5tki04gaJpZM4TDj9u> .
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/mkazhdan/PoissonRecon/issues/51#issuecomment-378106424, or mute the thread https://github.com/notifications/unsubscribe-auth/AkPxpdmZYoMQ4_CX2k8xy15dAc6qc1Dtks5tktvagaJpZM4TDj9u .
mkazhdan
commented
6 years ago Hi David,
Unfortunately, changing it there is dangerous. At the same time that the code computes the corner values, it also computes the marching cube indices for the cells / faces. You would need to make sure that the corner values and mc indices are consistent, otherwise you can get a "failed to close loop" error.
Two other ways to do this are:
-- Misha
On April 3, 2018 3:39:17 AM EDT, daviddrezner notifications@github.com wrote:
Thanks Misha I will be more than grateful to get your assistance with following,
My task is to overwrite the indicator function just following the poisson solver, with outband values (negtive values), so the final mesh will not pass through this voxels.
I've succeeded to overwrite some entire indicator function slice with a negative values (outband) and to split a mesh sphere. but I've having trouble to do it for only part of the nodes on this slice, there are following error: "[ERROR] Failed to close loop" _setIsoSurface() function.
drill down: my indicator function overwrite done in the _setSliceIsoCorners() function, set squareValues to negtive value just before MarchingSquares::GetIndex(squareValues, isoValue); function this action produced the error that I've mentioned.
Is there a way in the released code to mask specific voxels to prevent mesh passing through?
That's the source of my previous question regarding the order of the nodes within the slice, is the z-curve order is only of the active nodes? what is the relation to the voxel absolute location of that slice.
Best Regards, David
On Tue, Apr 3, 2018 at 5:18 AM, mkazhdan notifications@github.com wrote:
I’m not 100% sure I follow the question, but the nodes are ordered first by z-value (from slice to slice) and then by z-curve order (within a slice).
-- Misha
From: daviddrezner notifications@github.com Sent: Monday, April 2, 2018 9:53 AM To: mkazhdan/PoissonRecon PoissonRecon@noreply.github.com Cc: Subscribed subscribed@noreply.github.com Subject: [mkazhdan/PoissonRecon] poissonRecon nodes order on given slice (#51)
Hi Michael,
Can you please explain the order of the nodes for a given slice before MC algorithm. given to the poisson sphere ptCloud, I'm dumping nodes of the middle slice just before the MC (Marching Squares), having problem to understand the spatial order of sValues.mcIndices[i - sValues.sliceData.nodeOffset] hence corners indicator data.
Thanks, david
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub < https://github.com/mkazhdan/PoissonRecon/issues/51> , or mute the thread <
https://github.com/notifications/unsubscribe-auth/AKXc5RJqTmws90qzrxj8_ IITKJsUpjIkks5tki04gaJpZM4TDj9u> .
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub
https://github.com/mkazhdan/PoissonRecon/issues/51#issuecomment-378106424, or mute the thread
-- You are receiving this because you commented. Reply to this email directly or view it on GitHub: https://github.com/mkazhdan/PoissonRecon/issues/51#issuecomment-378157496
daviddrezner
commented
6 years ago Thanks Misha for all your help,
Great to hear regarding this capability in the coming version, Can you please estimate when this version with the additional points constrains would be released (days / weeks)?
Best Regards, David
On Tue, Apr 3, 2018 at 3:24 PM, mkazhdan notifications@github.com wrote:
Hi David,
Unfortunately, changing it there is dangerous. At the same time that the code computes the corner values, it also computes the marching cube indices for the cells / faces. You would need to make sure that the corner values and mc indices are consistent, otherwise you can get a "failed to close loop" error.
Two other ways to do this are:
- Extract the sampling of the indicator function on a regular grid, update values there, and then run standard marching cubes.
- Introduce additional point constraints to force the function to have prescribed values at specific locations. (I know this will be supported in the soon-to-be released version. Not sure if it's supported in the current version.)
-- Misha
On April 3, 2018 3:39:17 AM EDT, daviddrezner notifications@github.com wrote:
Thanks Misha I will be more than grateful to get your assistance with following,
My task is to overwrite the indicator function just following the poisson solver, with outband values (negtive values), so the final mesh will not pass through this voxels.
I've succeeded to overwrite some entire indicator function slice with a negative values (outband) and to split a mesh sphere. but I've having trouble to do it for only part of the nodes on this slice, there are following error: "[ERROR] Failed to close loop" _setIsoSurface() function.
drill down: my indicator function overwrite done in the _setSliceIsoCorners() function, set squareValues to negtive value just before MarchingSquares::GetIndex(squareValues, isoValue); function this action produced the error that I've mentioned.
Is there a way in the released code to mask specific voxels to prevent mesh passing through?
That's the source of my previous question regarding the order of the nodes within the slice, is the z-curve order is only of the active nodes? what is the relation to the voxel absolute location of that slice.
Best Regards, David
On Tue, Apr 3, 2018 at 5:18 AM, mkazhdan notifications@github.com wrote:
I’m not 100% sure I follow the question, but the nodes are ordered first by z-value (from slice to slice) and then by z-curve order (within a slice).
-- Misha
From: daviddrezner notifications@github.com Sent: Monday, April 2, 2018 9:53 AM To: mkazhdan/PoissonRecon PoissonRecon@noreply.github.com Cc: Subscribed subscribed@noreply.github.com Subject: [mkazhdan/PoissonRecon] poissonRecon nodes order on given slice (#51)
Hi Michael,
Can you please explain the order of the nodes for a given slice before MC algorithm. given to the poisson sphere ptCloud, I'm dumping nodes of the middle slice just before the MC (Marching Squares), having problem to understand the spatial order of sValues.mcIndices[i - sValues.sliceData.nodeOffset] hence corners indicator data.
Thanks, david
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub < https://github.com/mkazhdan/PoissonRecon/issues/51> , or mute the thread <
https://github.com/notifications/unsubscribe-auth/AKXc5RJqTmws90qzrxj8_ IITKJsUpjIkks5tki04gaJpZM4TDj9u> .
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub
https://github.com/mkazhdan/PoissonRecon/issues/51# issuecomment-378106424, or mute the thread
-- You are receiving this because you commented. Reply to this email directly or view it on GitHub: https://github.com/mkazhdan/PoissonRecon/issues/51#issuecomment-378157496
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/mkazhdan/PoissonRecon/issues/51#issuecomment-378231352, or mute the thread https://github.com/notifications/unsubscribe-auth/AkPxpQVD7wngZL89O1IkFim6rnqKE_bPks5tk2oagaJpZM4TDj9u .
daviddrezner
commented
6 years ago Hi Misha,
Regarding the additional point which constraints the function at specific locations, the format of this points would be as the input point cloud, but without normals and with indicator function value constrain? It will also force the octree to have the needed depth (up to maxdepth)?
regarding your option 1 " Extract the sampling of the indicator function on a regular grid, update values there, and then run standard marching cubes. " can you please share the extract function to a regular grid with me , or is it exist in the poissonRecon code?
I like to offer my help with this new release, I'm experienced alg eng. and programmer.
Best Regards, David
On Tue, Apr 3, 2018 at 4:46 PM, david drezner david@tetavi.com wrote:
Thanks Misha for all your help,
Great to hear regarding this capability in the coming version, Can you please estimate when this version with the additional points constrains would be released (days / weeks)?
Best Regards, David
On Tue, Apr 3, 2018 at 3:24 PM, mkazhdan notifications@github.com wrote:
Hi David,
Unfortunately, changing it there is dangerous. At the same time that the code computes the corner values, it also computes the marching cube indices for the cells / faces. You would need to make sure that the corner values and mc indices are consistent, otherwise you can get a "failed to close loop" error.
Two other ways to do this are:
- Extract the sampling of the indicator function on a regular grid, update values there, and then run standard marching cubes.
- Introduce additional point constraints to force the function to have prescribed values at specific locations. (I know this will be supported in the soon-to-be released version. Not sure if it's supported in the current version.)
-- Misha
On April 3, 2018 3:39:17 AM EDT, daviddrezner notifications@github.com wrote:
Thanks Misha I will be more than grateful to get your assistance with following,
My task is to overwrite the indicator function just following the poisson solver, with outband values (negtive values), so the final mesh will not pass through this voxels.
I've succeeded to overwrite some entire indicator function slice with a negative values (outband) and to split a mesh sphere. but I've having trouble to do it for only part of the nodes on this slice, there are following error: "[ERROR] Failed to close loop" _setIsoSurface() function.
drill down: my indicator function overwrite done in the _setSliceIsoCorners() function, set squareValues to negtive value just before MarchingSquares::GetIndex(squareValues, isoValue); function this action produced the error that I've mentioned.
Is there a way in the released code to mask specific voxels to prevent mesh passing through?
That's the source of my previous question regarding the order of the nodes within the slice, is the z-curve order is only of the active nodes? what is the relation to the voxel absolute location of that slice.
Best Regards, David
On Tue, Apr 3, 2018 at 5:18 AM, mkazhdan notifications@github.com wrote:
I’m not 100% sure I follow the question, but the nodes are ordered first by z-value (from slice to slice) and then by z-curve order (within a slice).
-- Misha
From: daviddrezner notifications@github.com Sent: Monday, April 2, 2018 9:53 AM To: mkazhdan/PoissonRecon PoissonRecon@noreply.github.com Cc: Subscribed subscribed@noreply.github.com Subject: [mkazhdan/PoissonRecon] poissonRecon nodes order on given slice (#51)
Hi Michael,
Can you please explain the order of the nodes for a given slice before MC algorithm. given to the poisson sphere ptCloud, I'm dumping nodes of the middle slice just before the MC (Marching Squares), having problem to understand the spatial order of sValues.mcIndices[i - sValues.sliceData.nodeOffset] hence corners indicator data.
Thanks, david
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub < https://github.com/mkazhdan/PoissonRecon/issues/51> , or mute the thread <
https://github.com/notifications/unsubscribe-auth/AKXc5RJqTmws90qzrxj8_ IITKJsUpjIkks5tki04gaJpZM4TDj9u> .
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub
https://github.com/mkazhdan/PoissonRecon/issues/51#issueco mment-378106424, or mute the thread
-- You are receiving this because you commented. Reply to this email directly or view it on GitHub: https://github.com/mkazhdan/PoissonRecon/issues/51#issuecom ment-378157496
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/mkazhdan/PoissonRecon/issues/51#issuecomment-378231352, or mute the thread https://github.com/notifications/unsubscribe-auth/AkPxpQVD7wngZL89O1IkFim6rnqKE_bPks5tk2oagaJpZM4TDj9u .
mkazhdan
commented
6 years ago The new code will support the functionality for point constraints, though getting a front-end will take a bit of work. (Points won't require normals. Points will take constraint values. And points will be added at the finest depth.)
For the regular grid, you can use the --voxel option. (If you need an implementation for surface extraction from a regular grid, you can grab the code from http://www.cs.jhu.edu/~misha/Code/IsoSurfaceExtraction/)
-- Misha
On April 4, 2018 8:27:03 AM EDT, daviddrezner notifications@github.com wrote:
Hi Misha,
Regarding the additional point which constraints the function at specific locations, the format of this points would be as the input point cloud, but without normals and with indicator function value constrain? It will also force the octree to have the needed depth (up to maxdepth)?
regarding your option 1 " Extract the sampling of the indicator function on a regular grid, update values there, and then run standard marching cubes. " can you please share the extract function to a regular grid with me , or is it exist in the poissonRecon code?
I like to offer my help with this new release, I'm experienced alg eng. and programmer.
Best Regards, David
On Tue, Apr 3, 2018 at 4:46 PM, david drezner david@tetavi.com wrote:
Thanks Misha for all your help,
Great to hear regarding this capability in the coming version, Can you please estimate when this version with the additional points constrains would be released (days / weeks)?
Best Regards, David
On Tue, Apr 3, 2018 at 3:24 PM, mkazhdan notifications@github.com wrote:
Hi David,
Unfortunately, changing it there is dangerous. At the same time that the code computes the corner values, it also computes the marching cube indices for the cells / faces. You would need to make sure that the corner values and mc indices are consistent, otherwise you can get a "failed to close loop" error.
Two other ways to do this are:
- Extract the sampling of the indicator function on a regular grid, update values there, and then run standard marching cubes.
- Introduce additional point constraints to force the function to have prescribed values at specific locations. (I know this will be supported in the soon-to-be released version. Not sure if it's supported in the current version.)
-- Misha
On April 3, 2018 3:39:17 AM EDT, daviddrezner notifications@github.com wrote:
Thanks Misha I will be more than grateful to get your assistance with following,
My task is to overwrite the indicator function just following the poisson solver, with outband values (negtive values), so the final mesh will not pass through this voxels.
I've succeeded to overwrite some entire indicator function slice with a negative values (outband) and to split a mesh sphere. but I've having trouble to do it for only part of the nodes on this slice, there are following error: "[ERROR] Failed to close loop" _setIsoSurface() function.
drill down: my indicator function overwrite done in the _setSliceIsoCorners() function, set squareValues to negtive value just before MarchingSquares::GetIndex(squareValues, isoValue); function this action produced the error that I've mentioned.
Is there a way in the released code to mask specific voxels to prevent mesh passing through?
That's the source of my previous question regarding the order of the nodes within the slice, is the z-curve order is only of the active nodes? what is the relation to the voxel absolute location of that slice.
Best Regards, David
On Tue, Apr 3, 2018 at 5:18 AM, mkazhdan notifications@github.com wrote:
I’m not 100% sure I follow the question, but the nodes are ordered first by z-value (from slice to slice) and then by z-curve order (within a slice).
-- Misha
From: daviddrezner notifications@github.com Sent: Monday, April 2, 2018 9:53 AM To: mkazhdan/PoissonRecon PoissonRecon@noreply.github.com Cc: Subscribed subscribed@noreply.github.com Subject: [mkazhdan/PoissonRecon] poissonRecon nodes order on given slice (#51)
Hi Michael,
Can you please explain the order of the nodes for a given slice before MC algorithm. given to the poisson sphere ptCloud, I'm dumping nodes of the middle slice just before the MC (Marching Squares), having problem to understand the spatial order of sValues.mcIndices[i - sValues.sliceData.nodeOffset] hence corners indicator data.
Thanks, david
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub < https://github.com/mkazhdan/PoissonRecon/issues/51> , or mute the thread <
https://github.com/notifications/unsubscribe-auth/AKXc5RJqTmws90qzrxj8_
IITKJsUpjIkks5tki04gaJpZM4TDj9u> .
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub
https://github.com/mkazhdan/PoissonRecon/issues/51#issueco mment-378106424, or mute the thread
-- You are receiving this because you commented. Reply to this email directly or view it on GitHub: https://github.com/mkazhdan/PoissonRecon/issues/51#issuecom ment-378157496
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub
https://github.com/mkazhdan/PoissonRecon/issues/51#issuecomment-378231352, or mute the thread
-- You are receiving this because you commented. Reply to this email directly or view it on GitHub: https://github.com/mkazhdan/PoissonRecon/issues/51#issuecomment-378582147
daviddrezner
commented
6 years ago Thanks a lot Misha for all your help
On Wed, Apr 4, 2018 at 5:09 PM, mkazhdan notifications@github.com wrote:
The new code will support the functionality for point constraints, though getting a front-end will take a bit of work. (Points won't require normals. Points will take constraint values. And points will be added at the finest depth.)
For the regular grid, you can use the --voxel option. (If you need an implementation for surface extraction from a regular grid, you can grab the code from http://www.cs.jhu.edu/~misha/Code/IsoSurfaceExtraction/)
-- Misha
On April 4, 2018 8:27:03 AM EDT, daviddrezner notifications@github.com wrote:
Hi Misha,
Regarding the additional point which constraints the function at specific locations, the format of this points would be as the input point cloud, but without normals and with indicator function value constrain? It will also force the octree to have the needed depth (up to maxdepth)?
regarding your option 1 " Extract the sampling of the indicator function on a regular grid, update values there, and then run standard marching cubes. " can you please share the extract function to a regular grid with me , or is it exist in the poissonRecon code?
I like to offer my help with this new release, I'm experienced alg eng. and programmer.
Best Regards, David
On Tue, Apr 3, 2018 at 4:46 PM, david drezner david@tetavi.com wrote:
Thanks Misha for all your help,
Great to hear regarding this capability in the coming version, Can you please estimate when this version with the additional points constrains would be released (days / weeks)?
Best Regards, David
On Tue, Apr 3, 2018 at 3:24 PM, mkazhdan notifications@github.com wrote:
Hi David,
Unfortunately, changing it there is dangerous. At the same time that the code computes the corner values, it also computes the marching cube indices for the cells / faces. You would need to make sure that the corner values and mc indices are consistent, otherwise you can get a "failed to close loop" error.
Two other ways to do this are:
- Extract the sampling of the indicator function on a regular grid, update values there, and then run standard marching cubes.
- Introduce additional point constraints to force the function to have prescribed values at specific locations. (I know this will be supported in the soon-to-be released version. Not sure if it's supported in the current version.)
-- Misha
On April 3, 2018 3:39:17 AM EDT, daviddrezner notifications@github.com wrote:
Thanks Misha I will be more than grateful to get your assistance with following,
My task is to overwrite the indicator function just following the poisson solver, with outband values (negtive values), so the final mesh will not pass through this voxels.
I've succeeded to overwrite some entire indicator function slice with a negative values (outband) and to split a mesh sphere. but I've having trouble to do it for only part of the nodes on this slice, there are following error: "[ERROR] Failed to close loop" _setIsoSurface() function.
drill down: my indicator function overwrite done in the _setSliceIsoCorners() function, set squareValues to negtive value just before MarchingSquares::GetIndex(squareValues, isoValue); function this action produced the error that I've mentioned.
Is there a way in the released code to mask specific voxels to prevent mesh passing through?
That's the source of my previous question regarding the order of the nodes within the slice, is the z-curve order is only of the active nodes? what is the relation to the voxel absolute location of that slice.
Best Regards, David
On Tue, Apr 3, 2018 at 5:18 AM, mkazhdan notifications@github.com wrote:
I’m not 100% sure I follow the question, but the nodes are ordered first by z-value (from slice to slice) and then by z-curve order (within a slice).
-- Misha
From: daviddrezner notifications@github.com Sent: Monday, April 2, 2018 9:53 AM To: mkazhdan/PoissonRecon PoissonRecon@noreply.github.com Cc: Subscribed subscribed@noreply.github.com Subject: [mkazhdan/PoissonRecon] poissonRecon nodes order on given slice (#51)
Hi Michael,
Can you please explain the order of the nodes for a given slice before MC algorithm. given to the poisson sphere ptCloud, I'm dumping nodes of the middle slice just before the MC (Marching Squares), having problem to understand the spatial order of sValues.mcIndices[i - sValues.sliceData.nodeOffset] hence corners indicator data.
Thanks, david
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub < https://github.com/mkazhdan/PoissonRecon/issues/51> , or mute the thread <
https://github.com/notifications/unsubscribe-auth/AKXc5RJqTmws90qzrxj8_
IITKJsUpjIkks5tki04gaJpZM4TDj9u> .
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub
https://github.com/mkazhdan/PoissonRecon/issues/51#issueco mment-378106424, or mute the thread
-- You are receiving this because you commented. Reply to this email directly or view it on GitHub: https://github.com/mkazhdan/PoissonRecon/issues/51#issuecom ment-378157496
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub
https://github.com/mkazhdan/PoissonRecon/issues/51# issuecomment-378231352, or mute the thread
-- You are receiving this because you commented. Reply to this email directly or view it on GitHub: https://github.com/mkazhdan/PoissonRecon/issues/51#issuecomment-378582147
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/mkazhdan/PoissonRecon/issues/51#issuecomment-378613051, or mute the thread https://github.com/notifications/unsubscribe-auth/AkPxpWzwBRGUSy4tcuVtXIb410ykYGg7ks5tlNQmgaJpZM4TDj9u .
Hi Michael,
Can you please explain the order of the nodes for a given slice before MC algorithm. given to the poisson sphere ptCloud, I'm dumping nodes of the middle slice just before the MC (Marching Squares), having problem to understand the spatial order of sValues.mcIndices[i - sValues.sliceData.nodeOffset] hence corners indicator data.
Thanks, david