Hacker News Reader

Comment by philsnow | Hacker News Reader

How has Blender succeeded here where Gimp failed? Is there some unsung heroic person who has imbued Blender with a sense of taste?

View on HN

Gimp refused to change their ways and still does, that’s why no one outside of a handful of enthusiasts use it. People have complained for years that their UI is too complicated, clunky, not intuitive, etc. and Gimp basically ignored all this feedback (to be fair there were attempts at changing this but ultimately not much changed).

Even just recently they’ve released a major 3.0 version and I thought “oh maybe they’ve finally addressed the UI issues” but nope, not much changed on that front, they still have stuff like “GEGL operation” front and center in the menus for basic functions.

Blender on the other hand reimagined their whole UI in version 2.8 and kept refining it later, even though there was friction in the community about it (since power users like the old UI) but thankfully they pulled it off and now they’re reaping the rewards for it.

> People have complained for years that their UI is too complicated

Isn't this a feature?

No. If a tool is too complicated or cumbersome to use then people will look for another one that’s better. Not everyone is an enthusiast that is willing to invest hours or weeks learning all the intricacies of an app if all I want to do is some light editing of a photo or very simple illustration drawing. I will choose a tool with better UX and easier workflow 100% of the time, and most people are like this.

So maybe the target audience isn't most people? I get that people like things to be simple, but some people love the depth of GiMP is a feature, not a bug.

I used to be a staunch critic of GIMP, recommend everyone use Krita instead and so on. But recently I've given the new 3.0 version a go and... it's quite an improvement if you adjust the time scales to decades rather than years. I've used GIMP first in 2000 (it was actually my fist "real" image editor, other than stuff like paintbrush) and a LOT of the finicky stuff has been removed or improved since then. Having do deal with per layer canvas boundaries? gone. Having a floating "ghost" layer whenever you made an operation and having to figure out how to merge that back? gone. Need to work with mask and group layers? You can do it, and it doesn't give a brain aneurysm trying to use anymore. You can even apply non destructive filters and whatnot to the group layers!

Having the UI spaghetti all over the screen because it's just a bunch of loose windows? the default is just a single window with panes. Want to condense that mess of tool icons into a single column of icons? you can do that! In fact, you can reorganize the UI in such a way that it's actually not that offensive (please GIMP devs, have better UI defaults!). Even tough it has a GEGL operation for non destructive filters I think that's just a holdover from how things were implemented initially because most of the "normal" and "color" filters are already non-destructive and operate (from the user point of view) just like the GEGL ones. They might have plans to further convert all other destructive filters and merge the GEGL operation ones into the normal filters, hopefully soon.

Oh and by the way, you can certainly push GIMP hard nowadays. I've been doing some testing editing a 32 float bpc (128 bpp!) 16K image with a bunch of crazy non-destructive filters stacked and it handled it like a champ with a few slowdowns here and there, and my 32 thread ryzen CPU was at times fully maxed (yay for multi threading) and the RAM management was quite impressive with things using up to ~50GB of memory (you do have to configure it let it use all that) and no memory leakages afaik, closing just the file and reopening again worked fine. Also, zero crashes! Can't say the same when pushing Krita hard. It's quite smooth compared to how things used to be.

Honestly the potential is all there, I just hope that the GIMP devs get a break from all the negativity they've been receiving since... time immemorial and perhaps the rate at which they can improve their software increases now with all the inglorious refactoring work they've been doing behind the scenes being done.

One example is that Blender embraced the use of pie menus, and Gimp ignored them. The Gimp team is just not open to outside ideas, and gets really annoyed when users of other tools request features from those tools that Gimp refuses to support, and reacts by digging in deeper and clinging to their bad design decisions out of frustration and spite. A really sad culture of NIH and 4Q2.

In general and with many other things, Gimp could have been so much better and easier to use, but they systematically and spitefully ignored their user's needs and requests about so many things, while Blender did just the opposite, listened to users and improve the user interface and mouse bindings, instead of being stubborn and parochial about it.

Ton is heroically tasteful but not unsung: the community rightfully adores him! (But not Autodesk.)

https://www.youtube.com/watch?v=eJwG-qt-sgk

Master Blender Pie Menus for Faster Workflow!

https://www.youtube.com/watch?v=f1fwxQi50FY

Enable Pie Menus in Blender 2.9 - Blender Tutorial

https://www.youtube.com/watch?v=H-7Hmpt9UmA

Create your own Pie Menu in Blender | Pie Menus

https://www.youtube.com/watch?v=41fXtvzJ3Ik

Blender - Pie Menu Editor

https://www.youtube.com/watch?v=B4DoESgzAfI

Extending Blender Pie Menus with Custom Operators using Python

https://www.youtube.com/watch?v=P8w-tswp0JI

What is 4Q2?

UR2Q2BSTR8! ;)

"Blender embraced the use of pie menus, and Gimp ignored them"

Instinctive recoil "cue xkcd https://xkcd.com/386/ somebody is wrong on the internet"

I had to take a little walk and think about it, What is a pie menu?

My instinctive first take was, as far as I know gimp has had pie menu since day one, at least as long as I have been using it since the late 90's 1.something. Need a menu item, right click, there is your menu. is this not topologically the same as a pie menu? Does a pie menu have to be radial? is radial any better than a list, I know I prefer a list, it does not look as cool but is much easier to read.

Of course, I've thought a lot about what a pie menu is.

An Empirical Comparison of Pie vs. Linear Menus, Presented at ACM CHI’88 Conference, Washington DC, 1988 (proves that they're significantly better than linear menus and explains why):

https://donhopkins.medium.com/an-empirical-comparison-of-pie...

Pie Menus: A 30 Year Retrospective. By Don Hopkins, Ground Up Software, May 15, 2018:

https://donhopkins.medium.com/pie-menus-936fed383ff1

What kind of pie menus does Gimp have, and for how long?

Or do you just mean "erzatz pie menus" as defined here (there's also a lot of stuff about software patents, FUD, AutoDesk, Alias, 3D Studio Max, and Blender there):

https://donhopkins.medium.com/pie-menu-fud-and-misconception...

>Ersatz Pie Menus

>Richard Stallman likes to classify an Emacs-like text editor that totally misses the point of Emacs by not having an extension language as an “Ersatz Emacs”.

>In the same sense, there are many “Ersatz Pie Menus” that may look like pie menus on the surface, but don’t actually track or feel like pie menus, or benefit from all of their advantages, because they aren’t carefully designed and implemented to optimize for Fitts’s Law by being based purely on the direction between stroke endpoints instead of the entire path, minimizing the distance to the targets, and maximizing the size of the targets. [...]

How do Gimp's pie menus compare with Blender's pie menus and pie menu editor that I linked to a demo of above, or Simon Schneegans's pie menus in Gnome Pie and Fly-Pie and Kandu? You'd think it would be easy for GIMP to adopt Simon's GTK open source pie menu work, which has been around for decades.

Gnome-Pie: Homepage of Gnome-Pie, the slick application launcher for Linux.

https://schneegans.github.io/gnome-pie

Introducing: Fly-Pie!

https://schneegans.github.io/news/2020/08/13/flypie

Show HN: Kando – A cross-platform pie menu for your desktop (kando.menu)

https://kando.menu/

HN Discussion:

https://news.ycombinator.com/item?id=42525290

I dug into this out of curiosity a few years ago. I thought 2.8 UI redesign was all thanks to Ton, but it turned out it wasn’t.

Ton was actually against a UI overhaul for decades. There’s a video where power-users and Ton were vocally dismissing the need for a better UI, using GIMP-like arguments. There was at least one failed UI redesign in the early 2010s which I think Ton was quite involved with.

But something happened, the nature of which I don’t know. Then, Ton became hands-off and allowed the UI overhaul to take place, which I recalled made actual UX designers work with engineers.

My memory is foggy and I don’t have sources readily available. But I’m hoping someone will fill in the gaps or correct my understanding of events long past.

There were 3 major open source graphics projects: - Gimp for 2D raster - Inkscape for 2D vector - Blender for 3D

At some time 10-20 years ago they all were powerful, but being hold back by a bad UI. Blender turned it all around with their UI overhaul some years ago. Inkscape seems to be doing some correct steps now toward that, although it is still hard to use (at least for new users). Gimp seem to be moving the slowest.

I find modern Inkscape almost identical to CorelDraw, which is what I learned in high school. Works well, and I'm considering moving some remaining CorelDraw graphics projects over to Inkscape soon.

Parts of the UI of Inkscape are very clunky. Like the tabs where live path effects live, thats awful. Also the zoom/pan on the canvas. Corel was very well polished if I remember correctly.

Taking the feedback about UI seriously was probably huge

Unfortunately it’s the reason indeed

I think the Open Movie projects really helped with that. They had artists and devs working closely together to polish up particular aspects of the application for each movie. The movies themselves also did a great job of raising the profile of Blender each time they were released.

Basically. This is the same problem with FreeCAD. They just don't care.

I think its more issue of not knowing how (efficiently). Consistent UX and polished UI is often about saying no to many ideas and people. This requires someone calling the shots which is hard for community projects.

FreeCAD btw got way better here in recent 1.0. You can almost feel that with some money they might get on the “Blender path”.

Not only feedback bout UI, but functionality in general. Looking at the changelog I found this issue [1] in which not only they fixed something objectively broken in the renderer, but also made workflows exploiting the broken behavior were still possible [2].

[1] https://projects.blender.org/blender/blender/pulls/133991 [2] https://projects.blender.org/blender/blender/pulls/136465

I think the real answer is in credits section of fund.blender.org. Sizeable contributions at scale of $1m as well as corporate user feedbacks started flowing into it ~2019.

I’d view that more as an effect of popularity rather than the cause.

Funding doesn’t tend to happen unless there’s already a lot of interest in a platform.

Khara(the Evangelion company) quit 3ds Max and transitioned to Blender around that timeframe, in collaboration with Blender developers. This seemed to coincide with Epic funding(anime people aren't rich people, broadly speaking, so I guess capitalisms happen in parallel and elsewhere).

0: https://web.archive.org/web/20190814061013/https://japanese....

Reply on news.ycombinator.com