Comment by bemusedthrow75 | Hacker News Reader

Comment by bemusedthrow75 | Hacker News Reader

Parent Full thread

> One area we’re “ahead of schedule” on is the much-requested non-destructive editing!

This is the purest of comedy. I applaud it. And I think a twenty-seven-year-younger version of me who first used adjustment layers in Photoshop in early 1997 [0] might find it even funnier to imagine this future.

This isn't "skating to where the puck was". This is "skating to where the rink was before they tore it down and built a shopping mall, that they tore down to build offices that they are now going to tear down to build a retirement home for GIMP developers".

[0] when, I note, GIMP already existed. OK so they got a bit distracted building a GUI toolkit, and we should not forget the significant value it added. It's the real transformational product of the entire endeavour.

But seriously: this wasn't voodoo, weird niche case or hypothetical stuff by the early 2000s, when they were still building a GUI image editing app that really relied on advanced users having Lisp knowledge. And they continued to kick it down the road for literal decades.

So they're finally doing it and they're still getting grief.

This is classic open source entitlement. If it's so important to you then why didn't you spend time writing patches in the last 20 years? This isn't a product you buy. Look at the commit history; for almost all of it GIMP is just a few people working on it; between 2 to 5 at any given time IIRC. All of them in their spare time AFAIK. These people are not your bitches to order about.

In full knowledge of the HN guidelines I feel confident in stating that you need to learn when to shut the fuck up. No really – "couldn't you have spent more of your free time on this a bit sooner" is just a shit attitude, and one that you can just keep to yourself.

I'm not even sure the first hack at this will remotely approach the full potential of non-destructive editing: https://gitlab.gnome.org/GNOME/gimp/-/issues/9244

Part of the problem GIMP has is that they chase perfect ideas at the expense of good ones.

Adobe don't. The Adobe codebase is cruft that has taken a similarly long time to shake out. Lightroom was seemingly built like a skunkworks project almost like a desktop webapp (scripting around a core renderer, SQLite, simpler mode-oriented GUI) to avoid the intransigence and friction of Photoshop's codebase.

But this is what GIMP is up against. Cruft that earns money, that is driving not just what users want but how they think about what they want.

For GIMP to be more successful it doesn't need live non-global filters like blurs, but it would have been enormously beneficial to the project if it had global adjustment layers, say, 15 years ago.

Then youtube would have been filled with tutorials, there would have been momentum around the project, some commercial support interest, etc.

Kicking the can down the road has consequences. This great Ondsel blog article about how difficult it is to merge major changes into an open source project goes into it in some detail:

https://ondsel.com/blog/freecad-topological-naming/

Some decisions need to be taken at the right time; if you don't, you need a whole separate kind of energy to get them done without stopping everyone else.

(The successful Affinity suite is a result of taking some right decisions back in 2010 or 2011)

Also here it is -- a moment of can-kicking in action, from that thread:

> Hey! I also think Adjustment Layers are really important for GIMP. However, Layer Effects will be simpler to start with since they only affect one layer.

As the person who said that, I don't 100% disagree - but also, I was just starting to look into non-destructive editing and had no idea of the scope. Since I didn't want to impact the much bigger kicked-can of GIMP 3.0, I thought having some form of NDE would be better than none at all - your "good solution" rather than "perfect solution".

(Also, it's not just non-destructive blur. All the color tools like Color Balance, Hue-Saturation, Levels, Brightness-Contrast, etc are internally GEGL filters and thus are now non-destructive. It's still early but we've already gotten positive feedback from users on how it's improved their workflow, or at least made it "less bad")

The good news is that now that we've worked on NDE, I have much better idea of how adjustment layers could be implemented and hope to implement those (or at least an intermediate step of adjustment layer groups) for 3.0.2. From other people's comments, it seems like that's the next really useful feature to have. :)

Reply on news.ycombinator.com