Getting Good Results (Page 2 of 2)

2nd try after tweaking again scanned the small cast metal truck pictured above

looking down at top of scan, can you see a truck? I dont

use some Talc powder to dull reflective areas. then blow/wipe it off when done

scan I am doing now using more auto settings on a Matchbox size APC tank no powder very little shine on it. what is the big arched thing sitting on top? its not part of model and no stray light like that. I made an enclosed booth with open top for room light to get it, black crate paper around scanner to keep laser from reflecting or getting out.

finished scan PLY in Meshlab. weird but easy to clean. I could live with this if it will also do it on smaller sized vehicles etc.

Camera view

took crown off but still got lots of gaps/holes and parts missed not sure how to replace those with Meshlab its not as easy to use as other apps nor as many tools

powdering model helped some but still looks like a wrinkled tarp is on item. settings used below. I may set noise removal to Med and see if that helps but doesnt seem likely.

This is what a scan of a matchbox I had came out like unedited.  I did not use any powder so the surface was shiny.  If I really wanted a good scan of this the powder would be needed and likely at least 4 scans from different angles.

Ply file in meshlab.

No I would never move the camera, there is only one spot that the camera is perfectly calibrated at. I mean that I would scan the model several more times. For example I would scan the model with it on its side to capture the top and bottom. Also move the car off center for a scan. This lets the camera see the object from different angles. Each will capture some portion of the object that another one didn't. I do as many scans as needed to get the majority of the object.

I find Meshlab easy to scroll and move the object around in.  I even often place my view inside the object looking out while I'm cleaning stray noise from the scan.  Maybe it's because I am using a mouse.

Meshlab has an align tool built in for merging the scans. Pirvan did an excellent tutorial on how to merge scans. I learned everything I know about scanning from this forum, I knew nothing when I received my scanner. It was a gift.

http://www.soliforum.com/topic/13039/al … -manually/

I didn't find any of these answers by searching, sometimes forum searches are difficult to use. This forum has less than 200 topics total. I went through them all and subscribed to the ones I though might be usefull. It's an hour spent that will save you many hours of frustration.

yeah same here on searching for answers. seems the search engine is not as good as some. I went through that tutorial but still have lots more questions about meshlab I cant find anywhere let alone in Meshlab. meshmixer is also hard to learn but has much more options not found in Meshlab. I mostly have been working with STL files and a couple related or converted to/from esy like Sketchup, Blender, etc. Sketchup 8 has tons of features for tweaking those files I have imported STL to it, modified several aspects then exported fixed STL  I have removed "hidden Geometry several layers deep, moved all around and into files to see things embedded in files either needing removed, repaired, rearranged, etc and can manipulate the file anyway needed. each program has pros and cons or different features. main trouble I find with Meshlab is moving object how I need to see it for repairs, then easy saving or undo of changes including un-delete like many programs do since they auto save dozens of moves so you dont cuss yourself for bad clicks like in the Atlas connection page when you meant to click a file to save and instead click delete on the last 2 hour scan done, that Pi dont have much spare memory so its understandable. but programs running on my PC have no excuse for being stingy with memory or tools. when I try to zoom in meshlab it only moves object towards camera view to about half way across screen while moving out of the gimble and is impossible to grab and move where needed due to limits within the screen. no tape measure, magnification by %, etc its confining.  other programs have better cleanup features, just not sure about things like merge multi scans. meshmixer does lots more with STL as to cleanup, repair, Mods etc. main feature only found on paid for programs is like removing parts of an STL that has multiple objects in 1 file or adding more which can be done in repitier host some or maybe with the other type STL besides just binary. since I mostly work with tiny files/objects I must have a way to inlarge them to see the details needing work. so far I am learning the Atlas Scanner unmodified is not high enough resolution for the models I use. wish somebody would have tried smaller/fine detailed objects than simple Troll dolls like you can buy at Dollar Tree. in the sand cast Nick Knacks or blow mold toys isles. need thinner laser lines and a focused Macro camera.

also each setting I select to improve fine details doubles scan time. just the settings I posted above take a little over 2 hours. heck I would be better off taking measurements of the item, learn better how to make CAD file and design the item from scratch like one guy whos models I print off Thingaverse. he does it in his lunch hour. now I see why Azerate sold his.

undo is like saving 50 dif steps of the model where saving over writes the file making any previous steps imposible to go back to.

I am good at micro detail work but afraid to mess with the camera getting glue off without ruining camera (popping it oc PCB) also have to disassemble partly to get camera back out of Atlas without ruining it also. but seems the only way to maybe achieve what I need to be able to scan tiny objects.

also need to see if I would be able to merge multi scans well enough that tiny with limited zoom. my eyes aint young anymore lol. those tiny wrinkle lines on my scans make a huge difference on misro size objects I print. I would say ATM the Atlas scanner resolution is at best down to 2mm (low poly) I need 0.001 min to be usable as high resolution as it states.

the out of focus camera wont see anything that small as anything but a wide blur. maybe after its been torn down and focused to proper Macro focal point which is anything less than 2-3 feet from lens. then yes I can see that. the "noise" I am seeing is well over 2.5mm which is a full decimal point thicker than that. my small models only vaguely resemble a ghostly fog of what the model is when scanned (and thats only due to color view on some models)in basic shape. most dont even show large parts like Wheels etc much less windows and door seams or grills etc.

just wish this had been reported before I bought one. its disappointing to learn after the fact. also some materials/components used in the kit are very poor quality like the power supply that refuses to stay plugged in and not loose connection most of the time within minutes of starting a scan. wasting my time with restarts. this is very frustrating since the Scanner has no battery backup or memory of scan that was interrupted. good power connection is imperative for operation of the device at all times. the push button power switch too has erratic connections at times.
photos of finished scans without color would likely have shown how little detail was actually captured.

kinda like some 3D file depositories I have searched through that have 3D color images of say a building but the STL file is only a rectangle box with no 3D details at all not even a doorway or windowsill. that is purposely misleading and not truly a 3D file. who buys a 3D printer to print bulk dimensional building materials available at a local lumber yard priced by the board foot?

those virtual 3D photo files are only for use by an architect for presentation apps on a screen in a board room meeting. not even a display model. and low poly 3D items have very little value in retail to support the cost of equipment used to 3D print same. only unique 1 off R&D/Prototype items or very fine detailed models near impossible to create using conventional methods without lots of tool & die work etc before hand like in mass production are mainly what 3D printing is proficient at.

I need to get the resolution down to at least same as SD2 0.05mm to be of any use. also can delays be reduced to speed scan times without making bad scans? I dont understand how some can do high res scans in minutes yet others including myself take at least 2 hours