-
-
Notifications
You must be signed in to change notification settings - Fork 986
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
New YOLO flow rate calibration #6479
Conversation
Hey Softfever! Excellent addition! Just one recommendation. Ideally we need 5-7 top layers to really show the flow rate variations plus a couple of layers of sparse infill to decouple the influence of any bed variation from propagating up. Basically would it be possible to make the models a little bit thicker? The Ellis print guide rectangles are just the right height :) |
Also maybe do the increments a bit smaller? Isn’t 0.05 a bit on the high side? What do you think? |
I agree with @igiannakas with his model size concerns. I, as well as a handful of other people I know in the community, always scales up the z on Orca's flow rate tests to get around the issue. It'd be nice to resolve that. |
Damn, I placed the decimal wrongly. |
@igiannakas @jeremytodd1
I noticed previously that two settings could ruin the calibration when the top/bottom thickness is set to, for example, 0.8-1mm, which squeezes the sparse layers. I don’t mind increasing the thickness a bit more to make it safer. However, 3mm, as Ellis used in his cube, seems a bit excessive since we have more settings to leverage here. I will make the following adjustments:
|
That should do it - personally I prefer a bit more top layers but equally on some occasions I’ve seen the Eli’s test showing me I’m ok with flow at 8 layers but then in the real world it is under extruding when doing 6 layers. So leaving it at 5-6 top layers may be just fine :) another thing I’ve noticed is layer height - at 0.2 I’ve been usually getting ok appearance but as soon as I drop to 0.16 or 0.12 it slightly under extrudes. May be worth considering a smaller LH but again will see through testing 🎉 finally 1 bottom layer may be pushing it a bit from an adhesion perspective. I’ve had the Eli’s cubes lifting their corners when using too few bottom layers but I see in this model the corners are rounded so this shouldn’t happen. Again something to test 😀 correction: 2 bottom layers should be good! |
Normal YOLO for 0.01 step Perfectionist YOLO for 0.005 step
I'm actually considering going in the opposite direction by increasing the layer height (LH) to amplify the visual difference between each testing blocks. However, I believe it's impractical to chase that level of precision in flow ratio. At the end of the day, there are too many external factors that can affect flow, such as filament diameter inconsistencies, off-center extruder gears, stepper motor imperfection, and chamber temperature differences that cause changes in plastic shrinkage. I personally only test down to a 0.01 resolution and then estimate a 0.00x number if necessary. |
With that I agree! Same here - 0.01 and then I may bump it up a bit if it’s somewhere in between. I have noticed that the Voron with the G2E is much better behaved and more consistent compared to the X1C when printing different layer heights. Maybe it’s just a limitation of the printer ;) also the Voron is more consistent filament to filament from the same material “family” compared to the X1 |
Now you tempted me to buy a Galileo 2 ;) |
@SoftFever I tried using German and French and in both cases the flowrate for every single specimen is |
* update flow rate calibration tests * more tweaks * add YOLO linear flow rate calibration * update name * revert line_width changes * Make it 2mm thick and change some text * Update YOLO test: Normal YOLO for 0.01 step Perfectionist YOLO for 0.005 step * add space
Hey, I wrote a comment here describing a method I used, but it seems on closer inspection what you've done is close enough that it covers my general usage of flow rate adjustment. I didn't see that the shapes had way more depth than the standard Bambu Studio calibration squares! Nice! |
@SoftFever Still not fixed. It took me half a day to read about it. |
I overlooked this message. |
It should be fixed now |
|
I think it would be nice to allow to configure the flow rate range, like we do with the temperature test or pa test |
When using Yolo calibration the top layer flow is still set to 1.00. Should not the top flow follow the flow so that you will se even better what flow is the correct one. Using 1.00 as top flow will make the difference way smaller when trying to compare what flow is the correct one. |
Effective top layer flow = object flow rate* filament flow rate * top layer flow rate. |
Some filaments seem tolerant of wide ranges of flow ratios, so occasionally you do end up with a whole plate of flow tiles that all look almost identical. I've noticed this most on filled filaments, especially wood. Presumably there are micro–cavitations between the filling and thermoplastic materials. It's normally recommended to use the highest flow which looks good in this case, which helps make sure you're packing the right amount of material in and there are no invisible gaps. |
The most obvious place I see overextrusion errors is when a monotonic layer is filled from two directions, causing the elephant's foot to progressively push ahead of the field as each line adds a bit more extra filament. How difficult would it be to fill the top layer from opposite corners, so they meet in the middle of the tile? Underextrusion and "perfect" extrusion should look the same but overextrusion becomes glaringly obvious, unless I'm missing something about the geometry of the lines? |
@SoftFever It looks like the Calibration Tutorial has not been updated to correctly reflect the new YOLO calibration options. |
Description
This test offers an alternative to the existing 2-pass flow rate calibration, requiring only a single print. The downside is that it relies on the initial value of the filament flow ratio not being too far off. However, given that Orca's current default profiles are quite reliable for most machines and users are familiar with the flow rate concept, this shouldn't be a problem anymore.
There are two YOLO versions:
The normal YOLO is highly recommended for everyone and is accurate enough for 99% of use cases, if not 100%.
Most people won't notice any difference in the test with a flow rate change of 0.005.
NOTE: In this new calibration test, the flow rate adjusts in a linear manner. For example, the -0.01 block indicates that the actual flow rate for this block is:
Screenshots/Recordings/Graphs
YOLO.flowrate.mp4