Saving in RC5 opening in 1.2.1 Multiplies Bump Map value by 4x!

[atelieryork]

Quite a nasty bug:

Open any corona scene, set a bump map value to 1.0 in a corona material, save it in RC5, then open it again in 1.2.1 and it will have changed to 4.0.

Do the same but at 2, it will come back as 8. 4, back at 16 and so on.

[atelieryork]

Interesting finding - it only does this if you have an actual map in the slot. If you set your bump to 1.0 but don't use a map, it will remain 1.0 when going from RC5 to 1.2.1. As soon as you use a map the x4 bug happens.

Oh god other things have changed too.

Absorbtion distance is changing... 0.2 changed to 0.289.

We tried everything here. So it's just the bump and absorbtion distance that are bugged.

[atelieryork]

I've uploaded the max file. No maps but you shouldn't need them. floor bump_RC5.max

To replicate, simply open this same file in 1.2.1 and those 2 values will change.

[romullus]

Not sure about absorbtion, but bump changing between 1.2 and 1.3 is not a bug. It's made for backward compatibility - bump strength was increased in 1.3.

[atelieryork]

Not sure about absorbtion, but bump changing between 1.2 and 1.3 is not a bug. It's made for backward compatibility - bump strength was increased in 1.3.

The value gets multiplied by 4 and definitely looks very different in RC5 than how it looked at 1 in 1.2, so if it's been done for backwards compatibility then it's very problematic. Surely if someone passes you a scene in 1.3 and you open in it in 1.2.1 then it should not change values. Because if they then work on the scene in 1.2.1, save it and send it back to the other person using 1.3 the values will be all wrong again.

[maru]

2. Backup your work to be 100% safe

Source: https://corona-renderer.com/blog/corona-1-3-rc1/

[pokoy]

Not sure about absorbtion, but bump changing between 1.2 and 1.3 is not a bug. It's made for backward compatibility - bump strength was increased in 1.3.

Wasn't this the case when going from 1.1 to 1.2? Is this changed once again for 1.3?

[karnak]

It would be helpful to keep track of these numeric changes between versions.
I'm thinking about a thread in the forum that keeps track of these multiplication factors.
Something like this:

Corona version	1.1	1.2	1.3
Refl. glossiness	1	?	?
Bump	1	?	4

[romullus]

Not sure about absorbtion, but bump changing between 1.2 and 1.3 is not a bug. It's made for backward compatibility - bump strength was increased in 1.3.

Wasn't this the case when going from 1.1 to 1.2? Is this changed once again for 1.3?

Maybe, i'm using dailies constantly, so for me it's very hard to keep track when something was changed :/

Anyway, did some quick test and it seems that indeed legacy check system gets broken once again. It affects not only bump but BRDF too. So it's urgently needs Ondra's attention.

[atelieryork]

Yes it really does need to be looked at.

This isn't really an issue about "backing up" and being careful. It's also about scene sharing. Imagine my colleague is working on 1.3 and I'm still in 1.2, or visa versa. That's going to be fairly common for the next months/year while people and offices upgrade to 1.3. Every single one of those projects is going to break...

[Ondra]

going back will trigger the change even though it is unnecessary, yes. We are aware of this, but it cannot be fixed now, because 1.2 is already out. In 1.3+ these situations will no longer happen - going back will not modify any values unnecessarily (but of course there could still be problems with missing maps etc)

[tallbox]

Quite a nasty bug:

Open any corona scene, set a bump map value to 1.0 in a corona material, save it in RC5, then open it again in 1.2.1 and it will have changed to 4.0.

Do the same but at 2, it will come back as 8. 4, back at 16 and so on.

As i said earlier i got same issue changing max versions and corona versions.