Hi Everyone,
For about a week I’ve been experiencing print quality issues after experimenting with slicer settings in Bambu Studio. I’ve since reverted my changes, inspected the nozzle, and cleaned it thoroughly. Despite that, the issue still persists: my top surfaces are not being laid down consistently. The infill looks fine, but once the top solid layers start, I see gaps/lines instead of a smooth surface.
Pic #2 looks like drying problems, and in general I would suspect the 25mm/s top surface speed vs the overall much faster rest of it. Low speed feed could begin to heat creep, you might be on the verge of clog as it takes its sweet time printing that section. Are you printing with the top glass off / door open?
Thanks for your suggestions! I do not have the top or door open. I never had to resort to that. I can turn up the cooling though? I changed it because ChatGTP said this was best to do. The 25mm/s top surface speed is because it did not came out consistent so I thought if I lower it it might work. The print takes up almost 16 hours.
I started to change the settings because I wanted to speed up. I only have the printer for 1 month so I am still learning.
The model is quite big so if you have other suggestions that might help I would really appreciate it.
The cooling fan in slicer settings is the one cooling the freshly printed model’s edge. The fan responsible for combatting heat creep is on the heatbrake of the nozzle, it’s already working at max.
I’d be careful using AI.
AI in some respects is an aggregator of the information it finds on the internet and there is no real way to validate what it spits out.
Google is running AI on search along with normal search and I have seen that Google AI be dead wrong about some things.
You may want to grab a cheap hygro-/thermometer and use one of the many holders on MakerWorld to place it in the heat chamber where the camera can see it.
In that manner, you can monitor your chamber temp and keep it at 35°C for (dry) PLA by proping the lid open. It is important regarding heat creep.
Models with fewer, smaller continuous surfaces, no ironing, etc can be printed well with chamber temps of 35-39°C.
I have nothing against printing faster but as you can see on picture one it skips a whole section for some reason… Anyway thanks for your advice, I will print with the door/lid open. See if this improve quality.
You are absolutely right. For about a week I was changing settings back and forth as chatgtp told me so. One time he told me to print 250mm for something and the next day he would say that was way too fast. So I cannot rely on that, therefore I came here to find other people who might had the same issue and hopefully resolved it.
Do the top layers look like that in the slicer preview?
You say you reverted your slicer changes but in your screen capture, there are a lot of changed slicer settings. I’d try reverting everything back to the defaults and if that fixes it, then add your changes back one-by-one till you discover which one caused the problem.
You really should not use Large Language Models like ChatGPT for anything where you cannot check the accuracy of the answers you get.
The term “AI” is so misleading. There is absolutely zero Intelligence behind these toys. All they do is string together words and phrases which might be connected to what you are asking about. The LLM itself has no concept of “correct”, “accurate”, “verifiable” or anything else like that. It generates text by guessing what you might like to hear and is incapable of determining if what it tells you is correct or not.
Let’s face it, the Internet is hardly a reliable source of information unless you already know enough about something to judge whether what you are reading is likely to be accurate or not.
ChatGPT and all the other LLMs say things which are not true or are totally made up all the time. As these LLMs are trained with what they can scrape from the Internet, the training data is full of false and incorrect information. Nobody checks this information before training the LLM - it would be impossible due to how much information there is.
It’s only a matter of time before the whole “AI” bubble bursts and a lot of investors who have no idea about how the technology works are going to end up losing everything.
This is called working as intended. Who cares about the lazy investor class anyways, lol, let em suffer for once. I don’t think anything short of actual open world war will stop their current power trip though, and even then, we all die to that so don’t worry about their precious assets. (99% mortality famously includes you lol).
.this was means to be uplifting
Didn’t quite get there. Anyways. Get skills and make stuff you’ll be fine
No the slicer preview looks perfect. And I understand why you would say that, I should have mentioned that I revert back my changes to the settings where previous prints did work. The standard bambu preset was already too fast in the beginning.Yesterday I started a new print with the z hop when retract at 0.2mm and kept the glass door open and this worked
Ah yes, I am using a 0.4 and I did 2 cold pulls that did not resolve it before. Yesterday I started a new print with the z hop when retract at 0.2mm and I calibrated my K factor and kept the glass door open and this worked for now.
You’re not wrong that LLMs aren’t “intelligent” in the human sense and yeah, they can spit out nonsense if you don’t fact-check. That’s real.
But calling them “toys” is kind of missing the point. They’re crazy good at speeding things up, brainstorming, writing drafts, summarizing, even coding. The value isn’t that they magically know the truth, it’s that they can crunch through patterns in language way faster than we can and give you a solid starting point.
The AI bubble probably will deflate from the hype, but the tech itself isn’t going anywhere. Same way people once said the internet was just a fad. Turns out it’s just another tool, powerful if you know how to use it, useless if you expect it to be a crystal ball.
With the help of AI I taught myself 3D design in a month. Was it always correct? Hell no. Did it frustrate me more times than I can count? Absolutely. But without it, I’d probably still be digging through random YouTube tutorials hoping to find the right answer.
& 