196 Reply by joan

(9 replies, posted in Ideas and feature requests)

I was thinking about this and another way could be to do it in the reverse direction with a "startlist". I think this could be useful for various scenarios. Even during training, since you could make a startlist for the various exercises you are about to do.

Something like this:

1. First you would create or export a CSV file containing the information of the sequence of videos that you are about to capture.
2. This CSV file needs to have column headers, for example it would look like this:

bib,name
2542,Tajay GAYLE
2206,Radek JUŠKA
2983,Marquis DENDY
2550,Carey MCLEOD
2200,Alejandro A. PARADA
2869,Simon EHAMMER
2449,Jeswin ALDRIN
2898,Thobias MONTLER
2410,Miltiadis TENTOGLOU
2552,Wayne PINNOCK
3041,William WILLIAMS
2170,Jianan WANG

3. From the moment this file is loaded, each new capture will consume one line of the file.
4. The column headers would be available to use in the file name templates. For example:

template: "longjump-%datetime_%bib-%name"

expansion: "longjump-20230904-1316_2542-Tajay GAYLE.mp4"

Now each time we record a video it would be saved with the right info directly in the filename.

Some considerations:
- If you record two cameras the file will probably have to be loaded in each screen separately.
- There should be a way to navigate within the list before recording, to fix mis-ordering and spurious recordings.
- It would be good to have the variables visible before the recording, for confirmation and in the context of training it could even be used as a cue to know what you are supposed to do next. At the very least there should be a button next to the filename to toggle between the template and its expansion, then it can be a global option where we show the future filename in big font on top of the capture screen (not part of the recording).

What do you think?

Go to forum: Ideas and feature requests

197 Reply by joan

(28 replies, posted in General)

Here is a list of the limitations and things I'm most likely to look at next

- In Kinogram mode, being able to change the time of each individual image, and change the font size of the labels.
- Counter and cadence tool.
- Camera motion compensation and related features.
- A way to export videos with more reasonable encoding parameters, for social media and whatnot.
- A number of videos people have shared end up cut short at the end for some reason.

Some of the machine vision camera plugins may not work with this version, I'll go over them as soon as I can.

Note: the translation coverage fell behind when I integrated the strings for the new features, you can help here:

https://hosted.weblate.org/projects/kinovea/#languages

Go to forum: General

198 Topic by joan

(28 replies, posted in General)

Post history

  • 2023-08-30: First release of 2023.1

  • 2023-11-12: Updated with links to 2023.1.1

  • 2024-05-16: Updated with links to Machine vision plugins

  • 2024-05-18: Updated with links to 2024.1.2

----
Hi everyone!

This release packs some major features and a lot of quality of life improvements.

There is still a list of issues and limitations I wanted to address but haven't had time to, but I don't want to delay this any further.

Note: at this point the release is not signed. I'm hoping to resolve this situation as fast as possible. (This means you may have to work around Windows smart screen to install it).

Links

    Kinovea-2023.1.2.exe (installer)
    Kinovea-2023.1.2.zip (self contained archive)

Camera plugins

    Kinovea.Basler-1.3.0.zip (Pylon 6.1)
    Kinovea.Baumer-1.1.0.zip (GAPI SDK 2.11)
    Kinovea.Daheng-1.1.0.zip (Daheng Imaging SDK 1.14)
    Kinovea.IDS-1.2.0.zip (IDS software suite 4.96, only compatible with uEye UI-xxx cameras, not the newer U3-xxx cameras).

In parenthesis the version of the vendor SDK or runtime against which it was built.
It may work on newer versions but if you have an issue try to get back to that specific version for compatibility purposes.
----

Here are some of the main updates (I know I probably say this at each important release but I intend to make a series of videos going into more details).

Kinogram mode

Interactive Kinogram creation, my favorite feature.

https://www.kinovea.org/screencaps/2023.1/2023.1-demo-kinogram.jpg

Multi-time chronometer

Each chronometer has multiple time sections and can show the cumulative time. Use this for split times or related measurements within the same video.

https://www.kinovea.org/screencaps/2023.1/2023.1-chronometer.jpg

Key images management

Each key image has its own color. There is a collection of "presets" that let you insert a key image with a custom name and color via a keyboard shortcut.

The key images can be reorganized and moved in time by dragging them onto the main timeline.

The comments have moved to a side panel.

https://www.kinovea.org/screencaps/2023.1/2023.1-demo-sidepanel.jpg

Spreadsheet export

For scientists and researchers: the spreadsheet export was completely re-written to be more exhaustive and target more modern versions of Calc and Excel. There is also a new JSON export in case you want to parse the data into your own programs and some tool-specific CSV exporters.

https://www.kinovea.org/screencaps/2023.1/2023.1-demo-spreadsheet-export.jpg

Document export

For coaches creating reports: it is now possible to export a document with the key images and the  attached comments, to prepare a report. The default format is Markdown but by  installing Pandoc (open source software) and telling Kinovea where it is  installed you can export to .docx or .odt. The export is kept simple on purpose so you can tweak the formatting and style in the  target application before creating your PDF.

https://www.kinovea.org/screencaps/2023.1/2023.1-demo-export-document.jpg

 
Quality of life improvements

  • Save vertically stacked videos and images.

  • Sort files by name, date or size in the explorer.

  • Support for undo/redo of drawing manipulation.

  • The drawings are rendered in the magnifier area.

  • The coordinate system can be added in Capture.

  • A new status bar in Capture indicates the state of recording.

There is more, including new experimental tools and changes I haven't mentioned, but I will leave this for later.

If you read until here, thanks for the support!

Go to forum: General

199 Reply by joan

(3 replies, posted in General)

I am locking this thread because it is a magnet for spam bots for some reason.

As for the original thread: I have implemented vertical stacking on export for images and videos.

Go to forum: General

200 Reply by joan

(3 replies, posted in General)

Quick update. I think a first tentative release for version 2023.1 is ready, I'll make a full post tomorrow with links.

I fixed some more bugs and updated the language strings for French, Spanish and Italian. These are the only languages I feel comfortable enough doing this with, but as I added/changed many menus and user interface the other translations generally fell behind. Help is greatly appreciated here -> https://hosted.weblate.org/engage/kinovea/

My code signing certificate expired and in the meantime the requirements have changed so apparently now I'll need to buy a more expensive certificate (+$300/year) plus a piece of hardware in order to sign the binaries, so the first release is not going to be signed. I'll provide the SHA1 hash on twitter for integrity checking, so at least anyone can verify themselves, but Windows will raise some warning messages during the installation.

There is a list of issues and important limitations that I haven't had time to address but I don't want to delay this any further.

Go to forum: General

201 Reply by joan

(1 replies, posted in Français)

Bonjour,
Pour l'instant non, les paramètres d'encodage sont en dur et sont optimisés pour la relecture image par image.
Cette limitation est assez haute dans la liste des priorités.

Go to forum: Français

202 Reply by joan

(3 replies, posted in General)

Hi, I should be able to post a first version over the week-end.

Go to forum: General

203 Reply by joan

(2 replies, posted in General)

OK I think I found the culprit and it should be fixed for the next version.

Go to forum: General

204 Reply by joan

(1 replies, posted in Bug reports)

Let's continue the discussion on github here: https://github.com/Kinovea/Kinovea/issues/121

Go to forum: Bug reports

205 Reply by joan

(5 replies, posted in Español)

Después de la grabación debería haber una miniatura del video creada de bajo de la pantalla de grabación. Puedes hacer clic en el nombre para cambiarlo (o clic derecha y "Cambiar nombre").

Go to forum: Español

206 Reply by joan

(5 replies, posted in Español)

La primera cosa es que solo funciona con las cámaras de tipo Machine Vision porque otras cámaras no tienen el nivel de control suficiente.

Si tienes una cámara de este tipo, puedes experimentar con reducir la resolución vertical, hasta que sea bajo de 16 pixeles (por ejemplo: 1024 × 4 px). Aconsejo ponerlo en lo mínimo posible. A partir de este momento, entrara en el modo escondido/experimental de finish line / photo-finish. En lugar de mostrar las imágenes individualmente, colocara las nuevas "bandas" bajo de las precedentes, continuamente. Con esa baja resolución, puedes aumentar la frecuencia de imágenes hasta 1000 o 2000 fps o más, según la cámara.

Nota: es solo activado bajando la resolución vertical, porque esas cámaras funcionan así. Bajar la resolución horizontal no permite aumentar la frecuencia, el sensor lee linea por linea de todas maneras. Hay que disminuir el numero de lineas. Entonces en la practica deberás poner la cámara físicamente tumbada al 90°, para que el efecto photo-finish sea el correcto.

En el fichero XML de preferencias, encontrarás un nodo `PhotofinishConfiguration`. Por el momento es el único modo de cambiar la configuración.

<PhotofinishConfiguration>
      <ThresholdHeight>16</ThresholdHeight>
      <ConsolidationHeight>2</ConsolidationHeight>
      <OutputHeight>1000</OutputHeight>
      <Waterfall>true</Waterfall>
      <WaterfallFlushHeight>100</WaterfallFlushHeight>
</PhotofinishConfiguration>

No sé exactamente si será útil para el salto de longitud pero sería interesante que más personas experimentaran con esto.

Go to forum: Español

207 Reply by joan

(2 replies, posted in General)

Most likely you are not doing anything wrong. It depends on the format but for some videos the frame accuracy doesn't work well and the working zone tends to reset itself. I have a number of examples I'm looking into it.

Go to forum: General

208 Reply by joan

(2 replies, posted in General)

Verifying account creation isn't sustainable, it's what I did for several years but now there is just too much volume. I just double checked the numbers, for the month of July alone I got 982 registration mail alerts. I'm not reading them. These are only the ones that manage to pass the captcha. About half of them go to a special honey pot page I set up but the rest is still too much. So we need to wait until they make their first move and reveal themselves.

The frustrating part is that the bots continue to attack even if it doesn't work. The website must be in spammers database somewhere and they just don't care if their attempts fail.

Some of the bots are also never posting but use the report button on random posts and put a link in the "reason" field, that doesn't make any sense, only a forum admin/mod can read this… I hope they don't start to do this more often.

Go to forum: General

209 Topic by joan

(2 replies, posted in General)

Hi everyone!

You may not realize this but the forum is constantly harassed by spam bots trying to post their links.

There are more than 10 new bots that register every single day on the forum and try to post messages. This has been going on for years and I have implemented many counter measures directly in the forum software code (which is unmaintained by the creators but I like its simplicity), so hopefully you shouldn't see any spam.

I catch some of them at registration time, others I catch when they fill their profile and signature, and others I catch when they post their first message. There is a secret sub-forum where they all end up together.

By the way if you are a human and your message is nowhere to be seen after you post, send me an email at the address at the bottom of the page so I can move it back into the main forum manually.

That being said since a few weeks there is a new form of attack I had never seen before, using AI / ChatGPT. The bot is making an actual answer by quoting the original question. The answer is looking normal and is vaguely on topic. But then they sneakily change the original question inside the quote to insert a link. Sometimes changing the font color so it is invisible. Sometimes they don't do it outright, they first post the answer and then come back a few days later and change the text in the quote. Very sneaky.

Obviously I would rather work on Kinovea than deal with this, and I don't want to restore the registration block where I had to manually validate every new registration, there is just too many bots to handle it manually. So if you see a suspiciously weird message don't hesitate to click the "report" link so I can have a look.

Thank you!
Joan.

Go to forum: General

210 Reply by joan

(1 replies, posted in General)

Not directly at the moment. The closest thing is to use OpenPose to generate .json pose files, then you can open the first one in the folder as an annotation file. It's not going to create trajectories though, each pose file is converted to a separate posture drawing on the corresponding frame.

Go to forum: General