Hi there.
My last scan.
220 mm panel. DMK UX236, k132, 12 mm computar. RMS 0.099..
13 scans, Fusion 3000-2.
Object is 355 mm long.
violin part
violin part
LG PF50, LG PF1500, RangeVision DIY: 2x DahengMer630, 2X12 and 2X16 mm 5Mp ZK lenses, RV turntable
Re: violin part
the top part:
LG PF50, LG PF1500, RangeVision DIY: 2x DahengMer630, 2X12 and 2X16 mm 5Mp ZK lenses, RV turntable
Re: violin part
These are OK, But I think you are reading the wrong RMS value. In the debug console there are a number of parameters that have RMS values. The one we are interested is an aggregate value. I'm not near my scanning computer so I can open the window and give you a better a better description right now. With an RMS value below .2 your scans should be aligning better than what you are showing in the pictures. This scan was done with a calibration that returned values in the .18 range.
µ
Re: violin part
Wood has not an homogeneus texture. Variation in refration index according to the fiber direction results in anisotropic behaviour to the incident light.
Using textures for alignement gave me not always good results.
Surface features helped me when textures gave ambiguous results.
RMS Is calculated as in the old discussions.
After "setup is inner corner".
I found this value to be minimum when panels are front tacong the camera. Tilting panels will results in higher RMS.
I wabt to improve two aspects:
-Vertical stripes
-Alignement
Will give a try to a cosmicar 8.5 and build 160 mm panels, in order to check the effect of relative dimensions of projector and camera pixels, and improve accuracy for this object size.
Will be grateful if somebody using non-hp hw will suggest me how to reduce stripes.
Using textures for alignement gave me not always good results.
Surface features helped me when textures gave ambiguous results.
RMS Is calculated as in the old discussions.
After "setup is inner corner".
I found this value to be minimum when panels are front tacong the camera. Tilting panels will results in higher RMS.
I wabt to improve two aspects:
-Vertical stripes
-Alignement
Will give a try to a cosmicar 8.5 and build 160 mm panels, in order to check the effect of relative dimensions of projector and camera pixels, and improve accuracy for this object size.
Will be grateful if somebody using non-hp hw will suggest me how to reduce stripes.
LG PF50, LG PF1500, RangeVision DIY: 2x DahengMer630, 2X12 and 2X16 mm 5Mp ZK lenses, RV turntable
Re: violin part
Hi all.
Some news: original David panels, and Stereo mode.
I found big improvements in alignement.
Now free align is 90% automatic.
fusion 2000, 2, 0.01%
This scroll belongs to a 200 years old violin. almost black in many details.
Very dark and somewhere reflective surfaces.
I played with projector brightness (calibrating at low brightness) and I'm quite happy something good came out.
I would like to try with gray panels.
Some news: original David panels, and Stereo mode.
I found big improvements in alignement.
Now free align is 90% automatic.
fusion 2000, 2, 0.01%
This scroll belongs to a 200 years old violin. almost black in many details.
Very dark and somewhere reflective surfaces.
I played with projector brightness (calibrating at low brightness) and I'm quite happy something good came out.
I would like to try with gray panels.
LG PF50, LG PF1500, RangeVision DIY: 2x DahengMer630, 2X12 and 2X16 mm 5Mp ZK lenses, RV turntable
Re: violin part
Here you have an example of the cropped texture.
Any suggestion pn how to improve image quality is appreciated!
(apart from putting something over the item! )
Any suggestion pn how to improve image quality is appreciated!
(apart from putting something over the item! )
LG PF50, LG PF1500, RangeVision DIY: 2x DahengMer630, 2X12 and 2X16 mm 5Mp ZK lenses, RV turntable
Re: violin part
very nice jobs madings. I'm glad to see your success in 3d scanning.
Re: violin part
Thanks bigbomber.
I had a good suggestion regarding stereo!
I had a good suggestion regarding stereo!
LG PF50, LG PF1500, RangeVision DIY: 2x DahengMer630, 2X12 and 2X16 mm 5Mp ZK lenses, RV turntable