Thanks, yes I use auto calculate and usually 0.5 multiplier. Using flush-to objects mostly slightly increases the effective flush a tiny bit. Removing the prime tower reduces it a little bit - so I guess it may end up resulting in having to push the multiplier up a little.
Will share the post processor once I have tested it out a few more times, plus worked out the best way of sharing python3 code in such a way that it is easy for others to install.
I have tried out some of the flush checking methods like the AMT_MW test models - bit I think it is probably going to be too complicated for to keep track of all of the different flush volumes for different colour combinations - so will probably just stick with auto calculate for now - because as long as you are flushing into useful objects an extra little bit of flushing isn’t necessarily a bad thing.
This is amazing. Just got the AMS added to my P1S and coloured my first print (turned Barry Bear into a panda) and I was disappointed to see my 90g print was going to make 180g of waste. I figured someone must have found a way to reduce that. I knew about flush-to-object so I had a couple of functional pieces I had waited to print for this reason but they only reduced the waste by 20 g or because they were either too thin or too short. Your experiment (which is how this read like Cmaneaonline said) was extremely helpful and got my waste down to 23g and 330g of functional. So although my filament usage increased to 22% my waste was reduced by 89%. And this was without using your table pieces or your python code (since I felt too inexperience to try it). I hope your research here becomes part of the Bambu Wiki as I think everyone could benefit from it.
Decided to have a go at individual colour swap purge lengths based on the AMT_MW models on maker world. Initial results showed definite benefits over auto calculate for some colour combinations - without a clear pattern.
As am currently away from printers - have used the time to create some automations on top of Bambu Studio using the Mac only ‘Keyboard Maestro’ app.
First macro is an easy one to automate the setting of all flush lengths to 50.
Next set of macros can then be used to store any newly calculated optimal lengths against the associated filament types and colours - so that they can then be easily re-applied whenever required.
It should also be possible to enhance these macros to do more generic enhancements to the build in auto calculate.
Initial ‘set lengths to 50’ macro shared below - if there is any interest in the others let me know.
I’ve got quite a few ideas for ‘flush into as colour doesn’t matter’ objects - such as
Large multi part functional items - designed with ‘flush-into’ specifically in mind, that can be built up over time - such as:
A flexible size table - made out of multi colour flush - that I am calling the ‘purge-table’.
Plus other big multi part things like shelving etc.
Xmas decorations - such as multi coloured balls etc -
Printer add ons - like AMS risers / AMS helpers / Scrapers etc
Storage - like Gridfinity etc.
Decorative things like Vases and 3DPB string art etc that are actually enhanced by the ‘random’ multi colouring
Big strong functional things - like the parts for the MPCNC for example
Finally got round to doing another multi colour print too tomorrow - this time of a BS hand coloured (my first using BS) Star Wars Corvette (from Thingiverse AirTeuteu Star Wars CR-90 Tantive IV split by AirTeuteu - Thingiverse) - flush will go down from 34g (30%) to about 1.25% (1%) using the ‘prime-table’ parts. It does push print time up by about 45 mins though.
Update 19/12 - First attempt at print failed - due to not spotting that top rows of engines needed support added - pretty happy with reprint though - and fairly small amount of flush waste.
Bit controversial I know - but I was wondering if I could be a bit more creative with the ‘purge-table’ legs as I am not doing enough multi colour prints at the moment to make much progress - so I was thinking about maybe incorporating some other ‘waste’ material in them in addition to the multi colour flush.
So printed with first ever Bambu Lab built in Benchy - have tried and failed a couple of times before - but it seemed to work very well on the cool plate on an X1C with the latest firmware ok. Quite noisy though I guess due to the high speed.
Not exactly sure how the structure the part - to make best use of the pretty strong structure of a benchy into the table structure - my first attempt is going to be something like this…
Thought I would have a go and see if I could do better with purge-table components.
With the model as downloaded from makerworld Bambu Seagull by marksolly - MakerWorld - total print time 25 hrs on a X1C - with 236g of flush vs 95g for the model - so about 250% flush.
If I change the flushing to auto calculate and reduce the multiplier to 0.5 (as per Teaching Tech YouTube) - then this brings the flush down to 132g. and print time down to 22.5 hrs
If I add in quite a few purge-table leg components for flush-into - I can get the flush total down to 5.7g - which is only a little over 5% - vs the original 250%. However doing this will increase the total print time quite a bit up to about 32hrs.
The next challenge after this is seeing what can be done with those massive supports - could most of the support also be replaced with useful objects too I wonder?
Correct - I am almost always increasing total plastic used - but using the saved waste plus additional plastic for something useful and that I was potentially going to print anyway.
In theory it would be possible to get the total additional plastic down lower with some additional different sized shapes or shapes that are wider or more complex at just the right heights where the big amounts of flush are needed.
Another way of getting the total amount used lower is by varying the amount of infill - so that there is more at the heights where flush is needed, rather than increasing the number of shapes. I guess I am sort of doing this by selecting shapes with top surfaces at different heights - so that they match where the most flush is.
I guess rather than building the top surface of the prime-table out of simple rectangles I could try and create it out of more complex interlocking shapes - that will therefore if printed in the right orientation may end up being more efficient in terms of less extra plastic used.
Original model 0.92g, with flush waste 13.19g - so over 1400% total filament 18.37g at 0.12 layer height
Decided to print a couple of ‘flush-into’ shoe horns with it - which normally take 26.22g at 0.2mm layer height.
New flush waste 0.8g, new total filament 50.64g - which is less than the total weight of the two ‘flush-into’ objects - bizarre must be to different layer heights or infill percentages or something.
Original model - which are tiny cruise ship propellors and motors (AZIPODS)
Flush-into object on its own - which is a shoe horn (tool for making it easier to get shoes on without undoing the laces) that I will put on makerworld.
Main print failed, and one of the shoe horns lifted off at one end due to bed adhesion problems - need to get some more Gold Textured plates as whilst I like the black ones - they are not great for long narrow or tiny models.
Made go use of the recently added ‘stop failed objects’ feature.
Anyway one of the shoe horns still worked ok and I am pretty please with it. Now on Makerworld. Colour doesn’t matter shoe horn
Did some stock taking of filament over the last few days - and decided that some of the older brittle PLA (some of it over 6 years old) and the short lengths really needs to be used up. So decided to use them to try and getting the first small version of the prime table built.
Took two approaches, purposely coloured one of the prime table segments to force colour swaps, then flushed-into all of the rest.
The biggest issue with this approach is that when each filament change happens for a spool that is almost run out and where the end has gone past the first stage feeder - then the only way for the printer to colour switch is to purge out all of the filament in the PTFE tube, which is quite time consuming and wasteful.
The 2nd issue is for brittle filament it occasionally snaps in the AMS or tubes so things have to be taken apart.
The 2nd approach I took was to just colour the Purge-Table parts all white, and set all of the colours in the AMS channels white, then just place loads of small lengths of filament or odd rolls in, and occasionally swap them around and cut the filament etc. This approach was more reliable, but the colour swaps don’t look as good as they are more widely spaced.
Also tried sport mode for the first time for both options and it worked pretty well.
All that remains now is for some joiners to be printed tomorrow, plus possibly a few misc size sections.
Then once I have a small prime table I can then just extend the size by adding extra sections as I start doing more multi colour printing.
First small complete version of purge table done.
AMS included for scale (plus proof that it is strong enough to be useful.
Design needs a few refinements - but overall I think it looks great.
Some better Prime Tower and Benchy holder component parts.
Test Prime tower holder hand modelled - but automated creation of a prime tower holder should be possible from analysing GCODE, or perhaps better still an STL of the prime tower that can then be used in Bambu Studio as a boolean object to make space for the prime tower to be incorporate into any ‘flush-into’ object.
Also heat press purchased so that I can try out creating some of the Flat sheets out of melted 3d printer waste demonstrated by ‘Teaching Tech’ YouTube a few weeks ago. Those flat sheets if they work could make up a good deal of the middle of a table top for example.
Good progress made today - firstly - in order to help manual AMS cut and swaps a bit easier made a nice little add on - to park filament when not needed. Will upgrade all AMS’s with this as I will also be a useful way to store the more brittle filament to stop it snapping overnight (or when away for a while) inside the AMS.
Also did a first test of the ‘Teaching Tech’ Heat Press melting filament in to flat sheets method - need to get a better frame and put more in next time - but pretty pleased with first test.
Had a little diversion while waiting for parts of ‘waste melting’ testing - printed out the BL marble run which is brilliant, plus now doing the Tourbillon.
Also made a start on a print queue management app (which ideally I think could be an addon to the slicer) - would really like a better way than plate naming to track in progress and completed prints - so started writing an app that processes the project 3mf files and gcode.3md files (if present) for prints a queue (which at present is just a folder on disk) - and extracts things like print completion status and predicted print times for each plate.
Should then be able to combine into the same queue the colour doesn’t matter files, plates and objects
Should hopefully then be useful for working out best plate to print next etc. - plus it may be better to track completion status in separate files within the app - rather than in the project 3mf files.
Made some further progress on queue manager.
Managed to analyse BS logs - and determine a way to track print starts automatically and link them back to the file and plate in the queuing system.
Can’t automatically determine successfully printed plates - but will try and do something with downloaded Timelapse’s triggering marking of successfully completed prints - will also mean Timelapse’s can be stored against the correct plates too which will be useful.
Needing to specify ‘latin-1’ encoding when reading the records.
Log records searched and used for tracking print starts from BS are those containing
‘load_projectfilename’ or ‘add_model_file_to_archive’ for collection of filename
‘on_send_print’ - to determine that a print has been sent.
‘export_plate_idx’ - after a print sent - for the plate number
For example plate 7 and 10 of this print currently in progress - and the printing of the two plates planned using the queue manager app, plus it also successfully picked up the fact that the prints of these too plates are now in progress (on 2 x X1Cs).