Author Topic: Updated to 2019.1.0 cant find my Collection materials from beta (Read 7380 times)

msjphotography · **Reply #16 on:** November 14, 2019, 04:26:57 pm

Baptiste, if it's not possible to run 2019.1 and 8.1 side by side, is there a way to preserve materials created in 2019.1 when uninstalling it and going back to 8.1, then returning to 2019.1?

iverlane · **Reply #18 on:** November 14, 2019, 05:54:05 pm

It would be nice if the installers did not delete older versions without asking you

Christopher Schubert · **Reply #19 on:** November 15, 2019, 02:29:20 pm

Quote from: Jeremie Noguer on November 14, 2019, 08:00:21 am

Unfortunately that's not a matter of spending money, the changes we made to the data system in Alchemist for the release are too deep to allow us to convert between them, we've tried but couldn't get satisfying conversion results.

People would be happy with something that takes in their old materials, creates new materials in the release version with the layers and filters they previously used, and sets the same values for the exposed parameters. For some people we're taking about hundreds of hours of rework this could prevent.

I've lead software development teams doing data extraction, transformation, and analysis in the financial and healthcare space for years, so I really do get it. I can't stop analyzing patient data from 15 years ago, despite the fact that healthcare data architectures are reinvented every couple of years. I can't discard account information if a client is moving from relational to object databases.

Migrating to a new schema or model can be complex, but it is RARELY impossible. We can see the v1 and v2 data models (or rather, infer) through the material JSON. This is not one of the cases where transformation is impossible.

If you need the data, regardless of what model it was previously in, you build a transformation. If you don't, it's either a matter of time or money.

seith · **Reply #20 on:** November 15, 2019, 03:38:34 pm

Quote from: Christopher Schubert on November 15, 2019, 02:29:20 pm

People would be happy with something that takes in their old materials, creates new materials in the release version with the layers and filters they previously used, and sets the same values for the exposed parameters. For some people we're taking about hundreds of hours of rework this could prevent.

I've lead software development teams doing data extraction, transformation, and analysis in the financial and healthcare space for years, so I really do get it. I can't stop analyzing patient data from 15 years ago, despite the fact that healthcare data architectures are reinvented every couple of years. I can't discard account information if a client is moving from relational to object databases.

Migrating to a new schema or model can be complex, but it is RARELY impossible. We can see the v1 and v2 data models (or rather, infer) through the material JSON. This is not one of the cases where transformation is impossible.

If you need the data, regardless of what model it was previously in, you build a transformation. If you don't, it's either a matter of time or money.

I hear what you say, Christopher. I've directed animation on several movies and I know what production's like. This is most definitely a matter of time AND money. It all comes down to someone in management having to decide that this was not a priority. The devs at Allegorithmic (yes, "Adobe") are not crazy. They didn't choose to give long-time users the finger. They likely knew this would not fly and yet they had to do it. And THAT's perhaps the most worrying part.

The leaders and developers of Alchemist are brilliant people, they've proved it. There's no way they would have misjudged the situation to such an extent. This looks a lot like a case of rushed release (I speak from experience).

All I can say is I've been playing with Alchemist 2019.1 and I can make the application crash at will (I've posted the bug report in the support forum), I've seen some of the beta issues are still present and you still cannot rename a collection once you created it. I mean... this is quite telling.

The way I understand it, Alchemist is still in beta but had to be released to coincide with whatever calendar was decided. It's still an amazing app and the Substance ecosystem remains excellent. I really want the best for Alchemist and the Substance team, so I hope this gets sorted out eventually.

seith · **Reply #22 on:** November 15, 2019, 05:34:37 pm

Quote from: Baptiste M on November 15, 2019, 05:04:39 pm

We're working for a few weeks on a prototype to transfer data from 0.8.1 to 2019.1.
To be fully transparent with you, here are the results and the limitations we found:

We succeeded to transfer the layer structure and images used in Import Image(s) Layers.

Here are the limitations we found (this list is not exhaustive):
- We can't transfer parameters of the Blend filter
- We can't recreate layers stack that contains custom filters
- We can't recreate layers stack that contains another layer stack as a single layer
- We loose the painted mask of the Clone Patch filter

What do you think about this level of data transfer and the limitations we found?

Thanks for your reply, Baptiste. I can only speak for myself of course, but that sounds quite acceptable to me (I never used custom filters, never used a layer stack as a single layer and only rarely used the clone patch filter). Assuming by "layer stack" you don't mean an existing sbrar file (because, well, materials all usually made of substance files).

Having to re-tweak the blends is kind of expected at this point. The most important thing would be to be able to get back all the layers (minus the exceptions you mentioned) pointing to the correct sbrar files that were used during the material creation. To me that's the most important thing: being able to get back all the ingredients of the original recipe, fully understanding that I'll have to "cook" them again.