Man vs Machine

Which camera works? Where can you buy an adequate laser? And more...

Re: Man vs Machine

Postby WalterMo » Thu Apr 30, 2009 6:59 pm

Hi Greg,
I really admire your work in electronics. Scanning and programming of course also. :wink: „Everything from scratch“ as you said.

Do you notice in the depth images the position from where Sven had completed the rest of the reference scan data by the just really scanned data? (If you are using the whole camera image for the repetition scan). Sometimes I can find at my scans a very small step or fine waves at this junction.

But with such a lack we must live. Important is that from now on with version 2.3.4 we can use nearly every worm gear. Even my very precise and therefore rather expensive one wasn't good enough for planeless scanning with the second last DAVID version. :(

Walter
User avatar
WalterMo
Moderator
 
Posts: 2249
Joined: Mon Apr 02, 2007 6:52 pm
Location: Braunschweig, Germany

Re: Man vs Machine

Postby Bongobat » Fri May 01, 2009 2:15 am

Hi Walter,

It is I who admire you as you are the pioneer in this area and have also been so kind as to teach me what you know. Oh And answer my numerous questions. Also admiration must go to Simon and Sven for programing this wonderful software and to all the forum members like Gunter and Mattia who have stuck around and posted their discoveries and scan results.

I did notice quite noticeable grooves or fine waves if my scan speed was too fast. Right now I am doing it painfully slowl but I am sure the problem is merely my worm drive. It has lots of play still and not really suitable for scanning but for testing my controller and sketches it is okay. As I am quite new to the planeless scanning today I had some discoveries of my own. First and foremost scanning bottom up is much better than top down as I was so use to doing.

Here is my first fusion of a true planeless scan. Up until this point all my tests were just repetitions with the same object and the corner in place so that the first enlightened pixel was the same. Today I wrote the software to my new controller so that I can program a scan and then repeat it many times. For these initial tests I wanted the controller to turn on the line at a point after stepper was moving then continue scanning. I removed the projector from the worm drive and used the DealExtreme laser. Although the controller is designed to be able to switch on an LED with a draw of 3A such as my backordered SSC P7. For first tests I use 4 buttons on my controller. First button starts recording and starts the movement of the worm drive, 2nd button starts the laser, 3rd button stops everyting, stores how many total steps was taken and when durring the scan the laser was turned on theb the stepper go back to the start position. The fourth button recalls the recorded scan. The small trimpot sets the speed of the scan. The scan is all done indepenend of the computer. I simply set up a corner, calibrate for that corner, use my controller to record a scan on that corner, , remove the corner and place the object down and scan it's various angles with the recall button.

Here is the result on the plaster bust I first scanned so long ago. Software and controller are working great but the worm drive is still a problem. I am very happy that the 8 different scans fit together pretty well. I did not smooth or interpolate as I left that to fusion. This was scanned faster than was needed and the ripple that I talked about above was present but for initial tests I am pleased. With the new David 234 cha cha cha being robust against gear backlash might give mne an excuse to hob my own worm gear.

Thanks,

Greg
Attachments
planelessfusion.jpg
Fusion result of plaster bust scanned without planes.
DealExtreme red line laser, Microsoft Lifecam HD-5000
User avatar
Bongobat
Moderator
 
Posts: 1002
Joined: Mon Apr 02, 2007 10:02 am
Location: Ohio, USA

Re: Man vs Machine

Postby Bongobat » Sat May 02, 2009 4:39 am

Hi,

I have run into a snag :oops: I have completed a planeless scan of a small dog statue but it seems the scale of each part is different and it will not align. The plaster bust was just rotated around it's center at 45 degree intervals durring scanning and went together okay. This time I placed the dog different places. Sometimes closer or farther from the camera/ laser to fill the camera screen. I was getting very good scans as I scanned this very slow but as you can see the different scans are a different size. I thought maybe something went wrong so I tried again but the same thing happened. Is this because I did not rotate it around it's center or could this be a problem with my hardware?

Any ideas?

EDIT; Okay I think I misunderstood how the first enlightened pixel worked. I thought you only had to have the laser to come on while within the reference scan view but from reading the wiki it looks like it has to come on at the exact pixel location in the reference and the repetition scan. This means when I moved the object the laser came on out of view but David thought it came on when the object caught the laser. Ah and now the talk about a physical trigger makes sense. :oops:

Greg
Attachments
planelessdog.jpg
Nice planeless Scan but...
different sizes.jpg
The scans are different sizes
Last edited by Bongobat on Sat May 02, 2009 10:36 am, edited 1 time in total.
DealExtreme red line laser, Microsoft Lifecam HD-5000
User avatar
Bongobat
Moderator
 
Posts: 1002
Joined: Mon Apr 02, 2007 10:02 am
Location: Ohio, USA

Re: Man vs Machine

Postby Bongobat » Sat May 02, 2009 10:27 am

Hi Walter,

I seek your advice once again.

I have ordered the SSC P7 and I want to switch it on the same way I switch the laser using the IRF540 to sink to ground. However. I dont want to use a high amp power supply to drive everything. Is it possible to sink the LED driver which has it's own seperate power supply with the IRF540 and still have the atmega8 microcontroller switch it? Basically tieing the ground of the LED driver board to the ground of the IRF540 which is in turn tied to the ground of the atmega8 power supply. I have thrown a quick drawing together. Sorry for the crudeness but I didnt want to hunt the stuff up in eagle ( there will be a resistor on the digital pin). Will this work? Im thinking not but not really sure why Possibly a difference in potenial in the grounds? :oops:

thanks,

Greg
Attachments
duel power.jpg
High Power LED switching with seperate power supply?
DealExtreme red line laser, Microsoft Lifecam HD-5000
User avatar
Bongobat
Moderator
 
Posts: 1002
Joined: Mon Apr 02, 2007 10:02 am
Location: Ohio, USA

Postby hal » Sat May 02, 2009 10:55 am

Hi greg and compliments for your work,
About your second-last message: yes, you are right: a wrong point of start for your Repetition scan give wrong results.
I think that happens because David, if read the first enlightened pixel in different position in comparison with Reference scan, is possible that it read the object as more big (ar smaller).
Sometimes just little difference are enough to obtain meshes of different scales. As you wrote, the only good system is the use of a trigger... but it can be a problem (we have deleted the physical obstacle of the panels and now we need an other physical reference... no sense, for me)
And an other problem is the lens and camera distortion: if we change de focus just a little or move back and forward teh camera (distance from object changes), appears some deformations.

MatM_distortion problems without panels.jpg

The upper side must be straight, but appears curved: this is a problem if I change too much the distance from camera and object after Reference scan: the lens distortion and the depth of field change, so change the way which David read the laser line.

I hope that in the future David can optimize this problem... :roll:

Bye, Mattia
User avatar
hal
Moderator
 
Posts: 1467
Joined: Thu Mar 01, 2007 4:08 pm
Location: Cavaria con Premezzo (VA), Italy.

Re: Man vs Machine

Postby Bongobat » Sun May 03, 2009 9:49 am

Hi,

Mattia, Thanks I understand now. I think I am done with first enlightened pixel. I have now programmed the atmega8 to take commands from David over the com port (rs232 to USB adaptor) and I set scan detection mode to 0. My major size problems dissapeared as I scanned this dog from all angles moving it back and forth and did not have any of the major problems as before. David control is just much easier and better I think I will desolder a few buttons. :wink:

So here is the fusion of the planeless scans of the dog statue:

thanks,

Greg
Attachments
planelessdogfusion.jpg
Fusion result of dog statue scanned without planes with the atmega8 controlled by David.
DealExtreme red line laser, Microsoft Lifecam HD-5000
User avatar
Bongobat
Moderator
 
Posts: 1002
Joined: Mon Apr 02, 2007 10:02 am
Location: Ohio, USA

Re: Man vs Machine

Postby WalterMo » Mon May 04, 2009 11:32 am

Hi Greg,
Sorry, but I was a few days off-forum.
To our question: Sorry again but at the moment I am not very familiar with the actual situation of your circuit board.
So let me recapitulate and correct me please: One output pin of the Atmega8 is used to switch your DealExtreme laser by means of an IRF540 FET to ground. And both have their own power supply with own transformers?
And what about the power LED Q5 in the modified projector? Was or could it already be controlled by the microcontroller? I remember we had talked about a constant current source with the LM317T.

And now you like to connect the new SSC P7 to the Atmega8.
I have taken a look on a German LED forum and found this values for it:
3.5V at 1.4A (385 lm)
4V at 2.8A (760 lm)
That means up to about 10W of heat must be conduct away!! Do you intend to use a „normal“ heatsink or better a Peltier element?
Do you want to use the same current driver as for the Q5 (a bit modified) to drive the new SSC P7 or will it come with an own driver? Then please give me a link to the site of driver and LED.

I think we will find a way to connect the new LED to the microcontroller. Care should be taken regarding the ground lines at this high currents. Anyway, using a relay will help in any case to separate the ground potentials.

By the way: I am glad to see (at your dog statue) that the new DAVID version 2.3.4 now perfectly works for the fusion at the planeless scanning.

Thanks and well done,

Walter
User avatar
WalterMo
Moderator
 
Posts: 2249
Joined: Mon Apr 02, 2007 6:52 pm
Location: Braunschweig, Germany

Re: Man vs Machine

Postby Bongobat » Mon May 04, 2009 12:25 pm

Hi Walter,

WalterMo wrote:So let me recapitulate and correct me please: One output pin of the Atmega8 is used to switch your DealExtreme laser by means of an IRF540 FET to ground. And both have their own power supply with own transformers?

No, the laser right now is using the same power source asthe atmega. It is a 7805 ( not drawn in my diagram) but the laser has it's own diodes to bring the voltage down to about 3.5 V. The 7805 as you know can not supply the SSC P7 adequately.
WalterMo wrote:And what about the power LED Q5 in the modified projector? Was or could it already be controlled by the microcontroller? I remember we had talked about a constant current source with the LM317T.

I am still using the LM317T as a driver for the Q5 but I have not tried to switch it with my board. I was going to start right off with the P7 and it's driver http://www.dealextreme.com/details.dx/sku.20330 but after I purchased the driver it go some bad reviews: only puts out 2 A and maybe overheating and cuts the current. If this driver doesnt work out I have already at hand an LM350T and was thinking I could use that like the LM317T to drive the P7.
WalterMo wrote:And now you like to connect the new SSC P7 to the Atmega8.
I have taken a look on a German LED forum and found this values for it:
3.5V at 1.4A (385 lm)
4V at 2.8A (760 lm)
That means up to about 10W of heat must be conduct away!! Do you intend to use a „normal“ heatsink or better a Peltier element?
Do you want to use the same current driver as for the Q5 (a bit modified) to drive the new SSC P7 or will it come with an own driver? Then please give me a link to the site of driver and LED.

When we start talking about Watts I dont know how hot that is till I burn myself on it :D I was hoping to use the same celeron heat sink. Maybe with a fan? The Q5 doesnt even make the heatsink warm. The only thing that gets hot with the Q5 is the LM317T and I was hoping to avoid that again by using the driver I mentioned above. the P7 I purchased is here: http://www.dealextreme.com/details.dx/sku.12721 It shows up as available but has been backordered for about 2 weeks now :roll:
WalterMo wrote:I think we will find a way to connect the new LED to the microcontroller. Care should be taken regarding the ground lines at this high currents. Anyway, using a relay will help in any case to separate the ground lines.

I agree that a relay would be the best solution and remembered you also mentioned that in PM. Would I need an optoisolator to prevent problems with the inductive load this creates? Since I am not really into the whole first enlightened pixel way of scanning anymore and since David tells my controller when to start through the COM feature, it knows everything it needs and there is no need to switch anything right now.

Thanks for all your help Walter :wink:

Okay now I have a new planeless test to show.
This was the angel also I scanned so long ago but now I rescan without planes and with texture grabs :wink: . Here is the the results rendered in zbrush. I have not done any texture touchup so you can see some magenta triangles where I was too lazy to get all those angles.
WalterMo wrote:By the way: I am glad to see (at your dog statue) that the new DAVID version 2.3.4 now perfectly works for the fusion at the planeless scanning.

and now with color texture too :wink:

David really is an amazing piece of Software :!:

Greg
Attachments
planelessangel.jpg
Fusion result of planeless scan of angel with textures
DealExtreme red line laser, Microsoft Lifecam HD-5000
User avatar
Bongobat
Moderator
 
Posts: 1002
Joined: Mon Apr 02, 2007 10:02 am
Location: Ohio, USA

Postby hal » Mon May 04, 2009 1:12 pm

Greg, you are the Man!... or better... the Machine!
Impressive results, really amazing :shock: :shock: :shock:

Bye, Mattia
User avatar
hal
Moderator
 
Posts: 1467
Joined: Thu Mar 01, 2007 4:08 pm
Location: Cavaria con Premezzo (VA), Italy.

Re: Man vs Machine

Postby WalterMo » Mon May 04, 2009 6:03 pm

Greg,
This ciruit should work to contol the SSC P7 by the Atmega8 via an IRF 540.
Unfortunately the voltage of a 5V power supply is not high enough for a sufficient LED current regulation. R1 can be used to take some heat from the LM350T.
What do you think?

Could you please enlighten your textured angel a bit? :idea: Again very well done.

Walter
Attachments
Atmega with LED.jpg
Microprocessor controlled Power LED
User avatar
WalterMo
Moderator
 
Posts: 2249
Joined: Mon Apr 02, 2007 6:52 pm
Location: Braunschweig, Germany

Re: Man vs Machine

Postby Bongobat » Mon May 04, 2009 8:11 pm

Walter, Thanks so much for the detailed schematic.

I will problaby use the LM350T as you have drawn but I think I am going to go with a relay rather than use the irf540. I notice you drew the power with 12V 2A but I was hoping to supply the P7 with 2.8 A The only power supply I have with enough juice is a power supply from an old computer. I think it says it will give 22 A on the 5 V and 16 A on the 12 V. So I suppose everything including the stepper can be driven by just the 12 V rail? Or maybe use the 12V for the motor and atmega8 and the 5V for the P7? What combination do you think would generate less heat? I was not wanting to use this but it looks like it may be my only choice the highest amp wall wart I have is for the cree and it is 1.5 A

Thanks again,

P.S The renderings were done with a zbrush shader called pearl cavity which I liked at the time but you are right after the rendering with GI it caused it to be a bit dark. I tend to like the Rembrandt style though. Here is another render but the model was decimated 3X in meshlab and automatic UVs were applied and then a texture transfer. I touched up a little bit but I notice I need to do more. Here is a render in zbrush with matcap white01 which is a little brighter.

Greg
Attachments
angeltouchup.jpg
matcap01 on meshlab decimated auto UV touched up model
DealExtreme red line laser, Microsoft Lifecam HD-5000
User avatar
Bongobat
Moderator
 
Posts: 1002
Joined: Mon Apr 02, 2007 10:02 am
Location: Ohio, USA

Re: Man vs Machine

Postby Khalid Khattak » Mon May 04, 2009 8:20 pm

Greg.. You are setting very tough target for us the newbies ;) ... I love to watch what u do..its simply impressive :) it looks really great...
PC: Dell Inspiron i3, 2.27GHz,4GB RAM,64 bit windows 7
CAMERA:
David CCD Mono USB
LASER:
David 5mW Green line laser
PROJECTOR:
3D LED Mini Projector 250lumens, Native 1024x600
[b]http://free3dscans.blogspot.com/
Khalid Khattak
 
Posts: 930
Joined: Thu Apr 23, 2009 1:30 pm
Location: Pakistan

Re: Man vs Machine

Postby WalterMo » Tue May 05, 2009 12:22 pm

Greg,
If you even intend to use the SSC P7 at 2.8A then it needs a drop-voltage of about 4V. And the LM350T also needs some volts for itself . Therefore you cannot connect it to the 5V of the PC power supply.
So take all from the 12V/16A.
I have never used such a computer power supply without a PC but for it's working in the switching mode I think it doesn't like to find no load at its outputs (during tests with a voltmeter). But I think you have some bulbs from cars lying around. :wink:
If you are using a relay instead of the IRF 540 to switch the LED there is no need to do it with opto-couplers. Connect a diode like the 1N4004 parallel in reverse-biasing at the coil. Or better is a faster one like the BYV27-200 or -400.

Walter
User avatar
WalterMo
Moderator
 
Posts: 2249
Joined: Mon Apr 02, 2007 6:52 pm
Location: Braunschweig, Germany

Re: Man vs Machine

Postby Bongobat » Wed May 06, 2009 8:18 am

Hi Walter,

I started looking for relays and I was thinking it would be cool if I could plug the relay right into the laser connector and use the irf540 to switch it but the only relay I have at hand is a 12 V for a car headlamp. It would not close with 5V and still wanted to draw 60 mA or so... seems a waste. Then I thought well I can use the unregulated 12 V since the irf540 will still just sink the relay to ground but I am still wasting current on the relay. So again I switch gears sorry. I think I have decided definitely to not use the relay and go with your last drawing but use the 12V rail of the PC power supply. This way everything has all the power it should ever need even if I choose to add another easydriver for a turntable. The real concern I had and the question I should have asked earlier is what kind of heat is the irf540 going to need to dissapate? Do I need a big heatsink for it as it is very close fit on the board. Also I know the LM350T is going to require a heatskink do you think a CPU heatsink would be enough? Like my Cree driver I would just throw that in a seperate box. I hate heat it is eventually going to destroy the entire universe :evil:

Also this way if I want to do outdoor scanning I can use those really neat rechargeable 12 V jumpstarters/power station http://www.harborfreight.com/cpi/ctaf/d ... mber=38391 or even just use the battery/cigarette lighter of a car :idea:

P.S The peltier is a really good idea but maybe for a future version.

Greg
DealExtreme red line laser, Microsoft Lifecam HD-5000
User avatar
Bongobat
Moderator
 
Posts: 1002
Joined: Mon Apr 02, 2007 10:02 am
Location: Ohio, USA

Re: Man vs Machine

Postby WalterMo » Wed May 06, 2009 9:53 am

Greg, what do you think how glad the SSC P7 would be having a cool peltier element just on its backside? It would radiate as never before. :D But a peltier also needs energy (regarding the global warming). :(

The heat which is to conduct away from the IRF 540 is very little. A rough calculation should be:
R DS on * I = 0.077Ohm * 2.8A = 0.22W
Important is to have a sufficient high voltage from the Atmega8 at the Gate of the FET. Have you ever measured this? It should be 4.5V or more.

The LM350T will become hotter. A small fan blowing on its heatsink will help. In this case a CPU heatsink should be sufficient. You can also try to take a part of the heat from it by using the resistor R1 in front of it. R1 can also be a bulb of a car. They act as a thermistor (PTC). Check it out.

Your shown 12V Jump Start device is a real power parcel. Sufficient for outdoor scanning for more than one night. :wink:

By the way: Driving a relay with 12V and a current of about 0.1A isn't an environmental sin. :wink:

Walter
User avatar
WalterMo
Moderator
 
Posts: 2249
Joined: Mon Apr 02, 2007 6:52 pm
Location: Braunschweig, Germany

PreviousNext

Return to Laser and Camera Hardware

Who is online

Users browsing this forum: No registered users and 2 guests