Sloppy focus errors inside Blender

Hi everybody,

I’m wondering how many people have been in the situation where they were pushing shortcuts on the keyboard while the mouse was hovering a different area of Blender than they thought or needed.

Example: “You are pushing the ‘T’ key to close the toolbar of the 3D view but it doesn’t work because your mouse pointer is over the timeline.”

Does it happen to you? Are you bothered by this?

Why am I asking this? The sloppy focus model is often used for advanced applications but even experienced users now and then run into the mentioned situation. I’m currently doing a research into the sloppy focus model by means of eye gaze tracking. It might be useful for Blender in the future, ie. the area you look at in Blender will be active instead of the area hovered by the mouse.



It is the way Blender works. The keyboard shortcuts are window sensitive. The only thing sloppy is your control of the mouse. The read your mind interface would be ideal, however.

Are you bothered by this?

All the time. I can’t tell you how many times I have been typing in code in the text window only to have the mouse drift over into the line number panel which except numeric text input to jump to a new line number. As soon as I type a number…BAM! The text window jumps and I am typing new code over top working old code. Ackh!

It’s one of the strengths of Blender which comes with a price. :slight_smile: Especially when coding in Blender indeed. Please vote on the poll so I can have some metrics of how much people this affects. Any comments on this are also much appreciated.



I’d have to agree with Atom, and add that this proposal sounds like a gross waste of resources and an unnecessary extra hardware requirement for users. Blender would need to access a compatible camera and then do facial recognition and motion tracking computations in the background just to get the window system to work properly. Instead, let me propose that people just learn how to pay attention to where their mouse is so their memory and cpu aren’t being hogged by eye tracking software when it could be handling their 3d work instead.

Don’t worry about the technical issues that’s not what the research is about. Just imagine that Blender would receive input of where you look. We are researching if sloppy focus can be enhanced that way. Blender is an example application where it could be an enhancement.

It’s still going to have to be computed somewhere. Either you’re going to need extra computational hardware to do the tracking, like a Kinect or something, or your computer is going to have to do it. If it requires extra hardware, that’s a pain for the user. If it uses your system hardware, I’d rather leave that devoted to a more complex CG scene. Technical issues matter. You can’t just handwave them away.

An eye tracker wouldn’t help much because if you touch type you look at the screen and that’s also where the mouse cursor is, if you don’t, then you’re looking at the keyboard (in case it has gone missing since last time you touched it) and eye tracker would point to the wrong place too.
Besides, eye tracker use with Blender has already been tested
While awesome, not very helpful.

LOL, that eye tracker is a funny joke of course.

In general if you use one hand on the mouse and one on the keyboard you must have run into the situation that your shortcut didn’t work because your mouse was in a wrong position by accident.

There must be more people that have run into this situation, especially power users. It’s not a big issue but it’s just a little interruption of your workflow. This is especially true when typing in the text editor for python scripts for example.

I answered the poll “It happens and it doesn’t bother me” because I see accidentally holding your mouse in the wrong place as an equivalent mistake to a typo. Sure, there are potential software fixes for fat-fingering your keyboard, but you’re still probably better offjust hitting backspace and dealing with the half a second it takes to correct the error yourself. Attempts to let the computer be your nanny are usually more annoying than useful. And you still haven’t properly addressed the system resource problems I’ve pointed out.

Well if you insist, just read the wikipedia article about eye tracking. There are a lot of applications for it. About system resources, it’s really none of our concern since that’s not the question nor is it this research. There are commercial hardware based eye trackers available which address that problem. What we are interested in is wether we can find any improvements when the computer knows what you are looking at. See for example this scientific paperabout improvements already researched (you’ll need access, sorry that’s the ‘paywalled science model’). Blender is just an example interface using the sloppy focus model which might benefit from this extra information. It’s not the computer that decides anything, like in your example, it’s what you look at that. The technical problems you describe are only relevant in the current commodity situation, in which you are right, however that’s not what this is about.

How can you honestly keep saying system resources are not a concern? 3D work is incredibly demanding on hardware, Blender more than some. As it is I often have to cut back on topology and turn off as many display features as possible just to get some scenes down to a level where I can even manipulate them easily in the viewport. I’d really like to be able to display more than I currently can. I don’t want to give up more of my precious cpu cycles and memory to some eye tracking software, when I’d rather use those resources to display textures better, or to handle a higher polycount mesh, or more particles, or just about anything other than eye tracking. Nor do I want to buy specialized eye tracking hardware just so my otherwise free software knows where to put my mouse. There’s a much easier way to tell it where to put my mouse: I just put it there. Done.

Tell me, please, why this is not a concern? I know it’s not what you’re researching, but frankly I don’t care what you’re researching. I care how this feature would benefit the program. Tell me how this feature would be worth the sacrifice of valuable system resources, and try to do it without sending me to a paywall.