search

alekssamos / cloudvision

Vision Bot NVDA addon
https://visionbot.ru/addon/
MIT License
9 stars 9 forks source link

As of NVDA 2023.1, the plugin is often buggy. #12

Open titet11 opened 6 months ago

titet11 commented 6 months ago

The following issues listed below will help you temporarily resolve some issues from NVDA 2019.3 to NVDA 2022.4. The solution will be provided after mentioning the issues below. When the plugin captures the image by pressing NVDA + CTRL + i keys, the plugin can recognize the text 2 to 3 times without problems.

However, after 2 or 3 times, the plugin simply no longer verbalizes the captured text.

2.

When I first want to read previously captured text using NVDA + CTRL +"I I", the plugin allows me to read the text without problems, but after I close the curtain to read the text, the plugin stops working completely.

That being said, the issues mentioned above were resolved as follows.

  1. At the time of writing this solution dated December 27, 2023, it is recommended to use NVDA 2022.4, as this is the latest version that can work without the bugs present as of NVDA 2023.1:

If you want to use an older version, the minimum and recommended version to use with this plugin is 2019.3:

  1. Now, you will be told which functions you need to keep active so that the plugin can work correctly using its keyboard shortcuts:

NVDA+5: activated:

NVDA + 6: disabled: NVDA + 7: disabled: NVDA + 8: disabled:

NVDA + m: On:

Also, remember to keep the following function activated in the navigation mode tab indicated in quotes below:

"Enable navigation mode on page load."

  1. If you encounter errors in NVDA versions 2019.3 or 2022.4, it is recommended to disable automatic updates for any plugin, as well as disable the following option for the IBMTTS plugin (indicated in quotes): "Always send your current settings":

Now, the problems that usually appear from version 2023.1 of NVDA are the following:

  1. Sometimes an error may occur when sending the analysis to the server:

  2. When you run recognition for the first time on some website or program, the text will be displayed complete, but if you change programs or interfaces, the text will be displayed incomplete and the scans will stop before the required time.

Once again I mention that this plugin is currently only recommended for use on NVDA 2019.3 and NVDA 2022.4.

The following comments were for the purpose of finding the solution I just mentioned, so you don't need to read them.

alekssamos commented 6 months ago

right.

titet11 commented 6 months ago

@alekssamos I already found the solution to all my problems with this plugin.

I simply proceeded to disable the NVDA+6 and NVDA+7 options.

On the other hand, you had told me to keep these 2 options activated, including NVDA+m, when the correct thing was to deactivate NVDA+6 and NVDA+7.

This resolved the incomplete texts and now text recognition works correctly.

The solutions indicated in previous comments were incorrect, so this solved all my problems.

Therefore, this has been resolved.

titet11 commented 6 months ago

additional note:

In newer versions of NVDA, such as 2023.3, 2023.2, or 2023.1, the text continues to shrink and the scan tends to take less time than it should, so it is recommended to use 2022.4.

titet11 commented 6 months ago

@alekssamos

Hello! I have updated the main comment of this problem with the solutions found. I hope it can be understandable.

Since there is still no fix for the bugs found from NVDA 2023.1 onwards, it is advisable to leave this issue open.