Subtxt features requests

Platform
1

Hi Jim,

Subtxt is great under Narrova. Here’s a couple-of-features request, primarily concerning synopses:

  • is it possible to only include prior moments when generating a synopsis, instead of pulling from the whole storytelling? As of now, it tends to include stuff in the synopses of the early scenes that only become known in the finale. (Example: the synopsis for moment 3 including an object A which is only acquire during moment 35) Maybe something like checking the current sequence of moments and their underlying subtxt storytelling? You’re the expert :slight_smile:
  • is it possible to completely empty the memory of a synopsis when regenerating it? As of now the previously generated synopsis is still clearly visible, which is unfortunate if I have since corrected stuff and want to regenerate from a clean slate.
  • Including the storyPOINTS in the basement is a great improvement. Unfortunately the synopses that are generated with included storyPOINTS tend to go very much off whack. Sort of writing a completely separate story on an own tangent/ disregarding lots of info that has been included in the subtxt events storytelling included in that specific moment, and filling in from other Moments (in the future).
  • Is it possible to choose to fill out the synopsis field myself (i.e. copy-paste my own synopsis into it)
  • I have uploaded codices for RAG (Retrieval Augmented Generation) in my subtxt story and storyform. Does subtxt check those when generating the synopsis? (My impression is not. Things that it clearly did check under Narrova, are forgotten or hallucinated wrongly.
  • Is it possible to generate a Report for a single Moment that includes all the subtxt info that has been lifted from the basement into that specific Moment? That makes it way easier for me to summarize etc in f.ex. NotebookLLM. It is also very handy to keep next to me when reworking a Moment.
  • Is it possible to make an instruction video/document, whatever, in which the precise usage of timing, and setting are discussed i.e. for best use practices. I find myself not knowing whether info present in the storytelling needs to be repeated in the setting/timing fields. My guess as of now is that only info present in that moment is included i.e. subtxt does not know which Moment comes before and after.
  • And, if I include timing/setting and use labels identical to those in an attached RAG-document, can subtxt then pick up the appropriate info from that document?
  • Would it be possible on a story-level basis to be able to replace the dramatica terms by ones of one’s own choosing. As it is now AI takes the terms literally, whereas sometimes I have my own specific intepretation of f.ex. logic, or emotion which is valid in that story specifically. It would greatly improve the applicability of Narrova generated material. (just as an example: logic might mean ‘Confucian ritual prescription’ and emotion ‘Daoist flow’ in a specific story)

Keep up the good work, it’s a great product!

Alison

2

What a great list. :star_struck: let me see if I can better understand what you’re asking for.

is it possible to only include prior moments when generating a synopsis, instead of pulling from the whole storytelling?

That’s what it should do. The generated Synopses (from within the Plotting section of Subtxt) should only be pulling from the Storybeats you have listed within the Moment. If that is not the case, let me know with examples (if possible), or if a secret story contact support with story id and the moment itself.

No spoilers are supposed to be in there!

  • is it possible to completely empty the memory of a synopsis when regenerating it? As of now the previously generated synopsis is still clearly visible, which is unfortunate if I have since corrected stuff and want to regenerate from a clean slate.

If what you’re saying is that after Generating Synopsis A you then Generate Synopsis B, and both still exist on-screen, then that is a bug. :lady_beetle: If you refresh the app and only Synopsis B is there, then the previous bug is definitely a bug. Let me know.

If you’re saying Generation B sounds an awful lot like Generation A, that’s a different issue…

  • Including the storyPOINTS in the basement is a great improvement. Unfortunately the synopses that are generated with included storyPOINTS tend to go very much off whack. Sort of writing a completely separate story on an own tangent/ disregarding lots of info that has been included in the subtxt events storytelling included in that specific moment, and filling in from other Moments (in the future).

That’s bizarre! I wonder–since you’re responsible for including Storypoints in this section!–the Points you are including, do they carry information about where that Point appears later in the story? The nice thing about having Storybeats there is that they’re only in that “act” whereas Storypoints often exist throughout different Acts.

In previous versions of Subtxt, you could only create one specific Storypoint per Perspective to keep things more manageable. I imagine now, as this isn’t the case anymore, and you can duplicate to your heart’s content, that there should be someway within Plotting to see the difference more readily between the two.

Unless I’m completely misreading your request on this one!

  • Is it possible to choose to fill out the synopsis field myself (i.e. copy-paste my own synopsis into it)

Yes, for sure. I will add that in.

  • I have uploaded codices for RAG (Retrieval Augmented Generation) in my subtxt story and storyform. Does subtxt check those when generating the synopsis? (My impression is not. Things that it clearly did check under Narrova, are forgotten or hallucinated wrongly.

By “codices” you mean…? No, it doesn’t include these but really should. I can add this as well (great idea!)

  • Is it possible to generate a Report for a single Moment that includes all the subtxt info that has been lifted from the basement into that specific Moment? That makes it way easier for me to summarize etc in f.ex. NotebookLLM. It is also very handy to keep next to me when reworking a Moment.

Another great idea. So more comprehensive report-oriented rather than “telling a story”, right?

  • Is it possible to make an instruction video/document, whatever, in which the precise usage of timing, and setting are discussed i.e. for best use practices. I find myself not knowing whether info present in the storytelling needs to be repeated in the setting/timing fields. My guess as of now is that only info present in that moment is included i.e. subtxt does not know which Moment comes before and after.

Yes, I can. Most of these features were installed to facilitate the Interactive side of Subtxt:

  • Setting describes where the current “scene” takes place.
  • Timing describes the relationship in time between this Moment and the one prior (e.g., “Three Days Later”)
  • Imperatives describe important points of Storytelling that might not be conveyed in the attached aspects (e.g., “Sara must tell John to never stop running here.”)
  • Players describes what “actors” should be included in the Moment.
  • Limit describes the number of “turns” a Playthrough can take before the system will intelligently end the Moment.

You could therefore, set up a Moment here with all that information filled out, click “Interactive”, and then play through the entire Moment in Narrova. RAG content saved to your Storyform and Story’s context will be drawn from if needed, and you can always jump out of character and steer the direction of the playthrough at any time. In this way, playing through your story becomes almost like a rehearsal where you get to be both actor and director (a writer’s greatest dream!).

When completed, you could always have Narrova generate a synopsis for you, or write out your experience in great detail for use in developing your story.

All that to say–yes, I can make a video.:sweat_smile:

3

Hah! So much to respond to, needed a second post:

  • Would it be possible on a story-level basis to be able to replace the dramatica terms by ones of one’s own choosing. As it is now AI takes the terms literally, whereas sometimes I have my own specific intepretation of f.ex. logic, or emotion which is valid in that story specifically. It would greatly improve the applicability of Narrova generated material. (just as an example: logic might mean ‘Confucian ritual prescription’ and emotion ‘Daoist flow’ in a specific story)

This is what Illustrations are for, and yes you can submit new ones like this. When interacting with an Illustration through the dropdown you simply enter your new version and tap Submit for Review. If accepted (it has to work semantically!) Subtxt will add it into the system.

Unfortunately, this part is currently under construction. It was one of the first things I developed way back when, and there are SO MANY better ways to get it done with greater accuracy that I need to update the whole method.

Hopefully, that answers everything. I really appreciate you taking the time to thoughtfully provide feedback and make suggestions, and I welcome new ones or follow-ups with open arms. :hugs:

Thanks!

4

Hi Jim,

Subtxt is a powerful tool, so thanks for clearing up my misunderstandings. I’ll whittle the following down to the necessary …

First – probably not the case, but worth briefly mentioning – as I work in time zone Amsterdam, I am often using various software in, what I call, the American update/roll-out time frame. All the apparent errors I encountered happened during one extended session during potentially that time frame. It has happened to me before with other software. Also, none of the potential errors occurred before I included storypoints in my moments. Hmm .

Okay, now for the still remaining issues one by on e:

- When I first generate a synopsis (A) and then discover I made some silly mistake, or just want to change it, I make the changes in the underlying attached events etc and regenerate (B). (A) disappears and is completely replaced by (B), which looks terribly similar to (A), so much so that it seems an underlying memory trace was just reactivated into (B)

- Yes, I did request the storypoints to be included in the basement for picking:blush:I will check my storypoints for unintended too-early-mentions in order to rule out the bleeding of the future into the present. I don’t think there were, but a check is warranted. I’ll get back to you when I have thoroughly checked any own potential misdoi ngs.

- By codices I mean separate documents uploaded to the Subtxt storyform for, at least, (i) characters, (ii) locations. But, you’ve got that one (and the others), so, th anks!

- @ single moment report: I think we’re understanding each other correctly. Everything attached to the same Moment in one report. For 10 different Moments, that would mean 10 different potential re ports.

- @ instruction for the setup options: The words are clear enough, but the underlying precedences are not. If I have a setting and timing in my attached storytelling, how does this combine with what I add in the Setting/Timing fields? Do I need to do the extra work to fill out the fields, or does subtxt scrape it from the info I already gave it? I use Imperatives to change the output style. Personally, I really don’t care for the internal telling and ‘as if-ing’ of any AI, so I just say keep it cinematic, no internal dialogue (etc). I also add in POV. That works quite nicely for me. And, yes, Interactive is great!

- @ own substitutes for DP terminology: Ah, thank you, I didn’t realize that yet, Now I do :blush:

Thanks for looking ,

Alison