I remember a time when 3D tutorials were only webpages with texts and screencaptures.
1 Like
I loove video tutorials but when it comes to quickly finding reference again I actually prefer these 
2 Likes
Having seen how UX research works in a big company: Alberto’s description sounds about right. Or at least, what I think should happen in order to get believable results. Because in the reality I’ve seen, the test groups are smaller. Which is still worth something – if six people try to use your software and none of them can figure out how to do it - well, that’s all the evidence you need that you have to change the software. But often the results are more ambiguous with a small group. But the main point Alberto is making here – that this is a large amount of work to do when there are a large number of things to test – is important. I would like that work to be done, but am not seeing a way to use the currently available resources to do it.
As a volunteer Blender developer, I am sometimes frustrated by questions about whether a new UI that I think of is good or bad. Because you can get a single person with a very loud voice repeating that a particular aspect of the UI is critically wrong, but I am left wondering: is it just that person, or is this really a widespread problem. An example of this was the migration of the normal tools from a GSoC project into Blender. We didn’t see a good way to include certain obscure operations in the UI, so left them out. One person was very loud about how that was a disaster, but when I solicited input from other on BlenderArtists, no-one else came forward with a use-case that needed those operations. It would have been nice to have a more systematic sampling method to really answer the question.
12 Likes
It might be possible to distribute the work a bit by coordinating with a few of the higher-tier corporate sponsors. It’d bias the testing groups, and it’d probably work better as a usability audit instead of an approval process for new features, but I think it could have a positive impact in the long run. Especially if the test groups can be balanced out by coordinating with universities.
That’d be a fair bit of leg work, though.
One time could be done, various each year no. No one is willing to interrupt their life or work, even if it is only a few days, every few months. It is always avoided as much as possible, often even when it is necessary, or it is done with very low samples that distort the result. Besides, it is very difficult to find people capable of doing it correctly and that their conclusions are useful. Well, it is a job absolutely dependent on the person who performs it and I assure you that it is normal that many people believes that they are able to do it but not.
Besides, although it is interesting to do it once on a small scale, adding it in the normal process of development would be a waste of time and resources that the foundation doesn’t have (neither do other companies much bigger) that would break the normal development of blender.
But we don’t want to see the obvious, the problem in blender has never been the lack of feedback, nor of ideas,… it has been that a design was decided in front of all that. And that can happen by having a large UX studio. It’s not going to change anything.
Generally it should not be that difficult to find out if the criticism valid or not, depending on if the person can come up with a reasonable arguments for it. In your case, as you said, no one was able to come up with a reasonable use case, so that pretty much solved it.
This doesn’t mean that finding answer to everything should default to some systematic sampling. I mean, most of the time the initial feedback will come mainly from the users who are actively interested in the feature, most likely because they really need it for what they do. That on itself serves as a good filter to get feedback from users who actually have an experience in the area of given feature.
Some more random polling of people can heavily distort the preferences because many of the people will answer either haphazardly or randomly, as they are not that interested in the feature and perhaps will never even intend to use it, so they may just answer in a spirit of “If I ever came across that feature, this is how I’d want it to work despite not really knowing much about it.”
So I think listening to people who actively seek out the development threads and show active interest in the feature even before it arrives to Blender is still the best way to go around it. It just requires a bit of effort to filter out voices which can’t really support their ideas, requests or criticism with feasible argument.
Also, maybe a bit controversially, I think that skill/experience of these feedback providers should work as a multiplier for a value of their feedback. From my personal experience, the more skilled one is, mainly in terms of either artistic or technical quality of their work, the better solutions they have, since they’ve had to solve a lot of problems and optimize lots of workflow to arrive to their high skill level.
That’s not it at all. For example, the case that Howard comments about the change was me and another artist, we were both the ones who did the whole GSOC process together with the programmer. The normal custom GSOC was complete, not a single feature needed by any artist or production was left out.
We spent several months polishing the UI, until it was in perfect use. Of course, once the GSOC was completed we assumed there would be no changes, so we mostly just walked away. At the time I noticed that the UI had been deleted from the normals in the codequest, which removed most of the functionality. The problem was that some tools needed a data input (directly enter the vector, multiply, add, …) and for that they need a panel. But as in UX it had been decided that the panels should disappear, the solution was to eliminate those functionalities.
I don’t know who was asked, but clearly they didn’t know the need of those tools in particular pipelines, like cartoon or some vegetation assets in games.
So the users are now unaware of the existence of those tools, which only exist now via script, by a decision of UX. And whoever needs them must use them via addon/script when they are hidden inside blender. Now is not “custom normals GSOC” is “weighted normals GSOC” because other tools have been lost.
There’s that big complicated UX process (usually done one time because it’s costly), and there’s also bad UX processes, with people that lack the insight into the product. It’s however also common to do really low level processes. I disagree with the scale that is needed. There are web cams and screen recordings to sample, which you’d have to use nowadays anyway. There is also a possiblity to filter this group, you’d obviously not need me in a sculpting feature request, doesn’t matter if I’m interested in UI or not. Testing “getting started in blender” would be a separate topic to me.
Say someone has an idea for a better panel in the render settings, and at the same time a couple of new features are coming. You first agree with devs and designers that’s it’s something you want at all, then you low level sketch it or make it into a quick prototype, at that point these get clicktested, before the actual implementation is started. Since people mention things in the backlog here from 2012 I don’t really think it’s going to slow anything down. At that point you know what you made is solid, and the treshold for changing something there should be higher.
I have no idea how hard it is to find volunteers, I don’t have the insights you have. I’m sure it would be possible to utilize some community figures for that. And maybe schools like mentioned.
I think it’d be a good framework to run things by. But you know this community better, I’m probably naive.
2 Likes
Howard, did you see this post? I would be interested in your opinion.
I think would be hard to find a sensible person who really answer “no” to your persistent question: “Should User Experience be a Core Value for Blender?”. I certainly say “yes” to that. In the end, if users don’t like and use the code one produces, the whole point of doing it in an Open Source environment kind of goes away. Though I am just a volunteer developer and without much say in what the leadership decides, I imagine they would not disagree that they want to provide a tool that is at once easy to use, delightful to use, and powerful. Those are of course hard things to achieve, and a bit of balance and tradeoff comes between the first two and the last.
I think the hesitation you are finding in getting a full-throated support for your proposition is a mixture of (1) the team already spent a fair amount of time on thinking about the user experience (in design sprints last year); and (2) a fear that to really do what you want to have done, properly, seems hard and would need a lot of people - hired and/or volunteer - to pull off.
Are you willing to volunteer your own time to work on this? You might convince people, for example, with some initial tests of different kinds of users, recording them as they go through typical tasks on Blender, and showing the lessons learned and potential improvements suggested by such testing. You don’t need anyone’s endorsement or permission to do this.
9 Likes
Thank you Howard for your considerate and thoughtful reply.
Yes, I would be happy to donate my time, assuming Blender management agrees to provide some strategic resources and priorities.
It would be nice to think a bottom up movement supporting design thinking, customer journey and UX design would succeed.
Unfortunately, my 35+ years as a designer tell me successful design is a top down, not bottom up endeavor. In successful engagements, management requests and approves such help. To not receive such support would be, I believe, a waste of my time and I would expect any such efforts to fail.
While Andrew Price is certainly no UX nor UI designer, and his previous efforts as such on Blender’s behalf were coarse and ill-started, the core concept behind it was admirable: make Blender easier to use.
If memory serves me correctly, other than mostly negative discussion, it went nowhere-- and provides an example of what happens when an outsider tries to change such things without proper support and approvals.
3 Likes
I think that was fear of producing an UI that did not fit 2.8 goals but was more inspired by 2.79.
They look at panel and thought that was possible to achieve same thing in F9 panel of Point to Target operator.
To me, UX of this operator is completely broken. Its modal keymap is still based on 2.7x keymap.
There are lots of conflicts in the use of the tool.
But what is most frustrating is that is a modal operator. It would be perfect to create an active tool version of it with a gizmo, pre-selection highlighting.
Presence of panel would completely fill 2.8 UI guidelines as an active tool.
Options would be clearly exposed when discovering the tools.
Instead of that, tools are in a sub-menu of mesh menu without really having a clue on what kind of selection using them.
Same kind of thing happened with tools from GSOC on vertex painting. There has been problem with the way to handle transparency. And GSOC UX workflow based on 2.79 was assuming that a Baking to Vertex Colors would be kept in 2.8. But it was removed with Blender Internal.
The GSOC was putting in place a whole workflow allowing to convert textures into vertex colors, vertex colors into weight maps and the reverse way was also possible.
Tools to do that are absent in master.
Clearly, here, there was a lack of coordination and understanding in making a correct transition from 2.79 UX to 2.8 UX.
Students were focused on 2.79 and user feedback. But at the end, mentors focusing on 2.8, did not really take the measure of the job done by students and broke UX.
They had to deal with general feedback on 2.8 and did not really give attention deserved to those tools or wanted to increase their workload.
I clearly would like to see active tools for normal editing and baking to vertex colors happening in a near future. Because that are holes in workflow. That part of UX is clearly broken without employment of non-official addons. And that is important.
That is not an elitist workflow. That is just allowing user to make adjustments, refinements from a low base of data.
You have few of polygons, need to deal with that. You can give them a great look by customizing normals.
You have only one texture allowed, need to deal with that. You can derivate masks from it using vertex color baking, adjusted in vertex paint mode. You can instantly make a weight map for modifiers from that derivative mask instead of from original texture.
We almost had that, by default in 2019. But no. And now, we probably will have to wait until 2022 ; that Pablo Dobarro finished his work on paint modes.
Well, actually the custom normals GSOC was well before we knew about the plans for the codequest and 2.8. It was designed with what we knew would be the UI of 2.8 in mind, which was the UI of blender 2.79 and its limitations.
We are not going to deceive ourselves, nobody really knew what was going to happen in the codequest, some people tried to mislead the users. People were told for months that the UX/UI would not change and in the first week of the codequest the T-shelf was erased, breaking half of the UI. The same thing happened with the wireframe, which was avoided to talk about the idea of eliminating it. So it was impossible to make a coherent design.
I also offered to redesign the tool UX for 2.8, even if it was just trying to save the tools and the modal operator was not changed. But since UI was ignored, they wanted the menu and what didn’t work in the menu was deleted.
Oh god … yes. I still really don’t understand how eliminating wireframe was ever considered a serious point to discuss. And to this day the wireframe overlay slider instead of fading out the wireframe overlay (useful) just drops every other line in the viewport (makes no sense at all).
1 Like
The Blender model of the Swiss Army Knife has collected way too many blades to be usable.
Something that would improve my user experience is the ability to add or remove sections of the software as my workflow changes, to keep them from cluttering the RAM and screen space and menus.
- The video editing module
- The “grease pencil” 2D
- The animation rendering
1 Like
Imagine the vision of a Blender that is a schools first choice to teach 3D because it is so simple for students to start and learn.
I’ve taught heavy duty desktop publishing software and it’s a matter of laying out a sequence of classes to show what you want the user to learn in small doses rather than creating “Adobe Publisher Lite” and then letting them flounder when they have to make the switch to a pro-level program.
That “donut series” is a very well done example of how to lead someone into Blender with a controlled series of actions.
That anyone wanting to update their kitchen, or fiddle in their workshop would see as the obvious and simple choice.
You want to aim them at the obvious and simple choices already out there, like Sketchup, TinkerCad or any of the low-powered 3D home design packages.
1 Like
That’s the idea behind Application Templates, you can kind of configure yoiur UI to your needs and taste.
You can sort of do that already in a simpler way with Workspaces, but with App Templates you can also enable / disable addons, default blend file and other things.
Are you saying that you could take Blender code, as it exists, and remove or hide the parts you don’t want … then as the students increase in ability, unlock the complexity?
That would be totally awesome!
Adding: Now that I know they exist, they sound like a very good thing. Much like a word processor’s template and defined styles …
Do I have to learn Python to make these things, or is there a way to avoid coding?
I can see that, and can respect your decision to wait until asked. I can also see that there is a bit of a dilemma: how can Blender management know that you are capable and compatible? You have talked about your resume, which is certainly a good indicator, but there is more than just a resume to consider when trying to decide if a particular person will fit in with and provide value to another particular group.
A bottom-up approach to effecting changes in something like this risks this: coming across as “just someone’s opinion”. And it is hard to draw out of a sea of conflicting opinions what the best course of action is. Which is why an approach that has believable data might succeed where previous ones have failed.
7 Likes
Not exactly locking things, I don’t think you can actually LOCK any part of Blender per se, but you can configure the UI in an alternative way leaving just the things you may need.
Think about this like configuring the UI for different departments in a production, where you want them to focus in their tasks.
Here is the info: