1,636

(8 replies, posted in General)

Thanks for all the propositions !

About coining a new word for the function:
The good part is that it creates a recognizable name, and that it does not need to be translated between all languages.
The cons is that it is not obvious what the function will do before the user clicks on the menu or check the help file.

In this, way, I think it's best to have something that is the most descriptive possible. I would like the program to speak the same language as the user so there's no surprises.

We may use a brand new name when the feature to name is completely innovative and doesn't correspond to anything.

I'm also a fan of the early history of cinema smile.
And that is why Chronophotography is already reserved for a future feature. It was a tremendous technique to decompose motion used by Muybridge and Etienne-Jules Marey in the 1880's.
I find this incredible how these guys were studying motion from photography more than a hundred years ago. They are totally forgotten in my opinion. A century later, we should have hardware that does that kind of stuff.

In trying to get as close as possible to describing the feature in actual words, I think "Overview" is pretty much spot on.
I have switched to that in the just-posted 0.8.6 experimental version, to try it out in the light.
Let me know if you have strong objections against it, otherwise I guess it's a keeper.

1,637

(4 replies, posted in Français)

Version expérimentale : elle a besoin de vos retour d'expérience pour s'améliorer !
Veuillez considérer cette version comme une « Release Candidate »

L'installeur est dispo ici :  Lien retiré. Voir version 0.8.7.
1. Suivi
- Nouvelle amélioration de l'algo, les résultats devraient être plus robustes. (fixe le problème de « template update » qui faisait que le suivi changeait de cible petit à petit.)
- Le code sous-jacent pour le matching est maintenant délégué à OpenCV, -> gain de vitesse important !
- La fenêtre de recherche est maintenant proportionnelle à la taille de l'image.

Comme d'habitude, décrivez vos scénarios de suivi qui échouent.

2. Manuel d'utilisation
La version anglaise du manuel a été mise à jour, et est en test dans cette version.
Pour le voir il faut basculer la langue en anglais puis faire F1 ou Help > Kinovea Help…
Vous pouvez aider à sa relecture :
Si vous trouvez des fautes de frappes, de grammaire, des passages obscurs ou manquants, vous pouvez les rapporter ici; dans le sujet du forum anglais; ou mieux: les fixer vous-même. Il suffit d'aller sur cette page du wiki et d'éditer le sujet correspondant.

-
En dehors de ça, les bugs suivants on été résolus : m175, m176, m177, m178, m179.

Capture : La séquence de test du Billard.  N'avait jamais passé le test jusqu'à maintenant big_smile.

http://www.kinovea.org/screencaps/0.8.x/tracking.jpg

1,638

(7 replies, posted in General)

This is an experimental version : it needs your feedback to improve itself.
Please consider this version as a Release Candidate.

The installer is available here: link removed. Please check version 0.8.7.

1. Tracking
- The algorithm was improved again, results should be more robust. (fix for the "template update" problem, when the tracker was slowly changing target over time.)
- The underlying code for the matching is now delegated to OpenCV, which provides a great speed bump !
- The search window is now proportional to the image size.

As always, please detail your failing tracking scenarios.

2. User manual
- A candidate version for the updated manual is embedded in this version, (Hit F1 or Help > Kinovea Help…).
- Please help in proofreading it:
If you find typos, grammar issues, missing bits, obscure paragraphs, etc. you can report it here, or better yet : fix them. Just go to this wiki page and edit the corresponding topic.

-
Aside of that, the following bugs were fixed : m175, m176, m177, m178, m179.

Mandatory snapshot : the Snooker tracking test sequence. It used to fail miserably, now both balls are tracked all the way big_smile.

http://www.kinovea.org/screencaps/0.8.x/tracking.jpg

1,639

(6 replies, posted in General)

I did not realize that Wine was supporting the .NET Framework…
I'll also try to experiment with that when time permits.
Thanks for the initiative ! Keep us posted if you make some progress smile

1,640

(7 replies, posted in General)

Hi,
It is on the suggestion list smile
One other request was to be able to rotate by 90°. Technically I think it will be the same complexity to code the rotation by arbitrary angle like you suggest or by fixed 90° steps.

The end result might be of slightly less quality than the original because for arbitrary angle we will need to interpolate the pixel values.
For the user interface, I was thinking of a new dialog box where you can change the angle with a slider and validate. (Other ideas welcome)
Bug m181 will also need to be fixed during the implementation of this function.

This will not be part of the next official release, maybe in the one after, depending on other prioritized stuff.

1,641

(2 replies, posted in Français)

Bonjour,
Le tracking qui est actuellement dans la version de développement est meilleur en termes de qualité et de vitesse de traitement. (bien meilleur à mon avis mais je vous laisserai juger smile)

Le chiffre correspond à une donnée interne de l'algo, c'est le coefficient de corrélation entre la zone de l'objet (rectangle interne) de l'image précédente et celle de l'image courante. En gros cela mesure la réussite du tracking pour l'image en cours. La valeur va de +1 à -1, mais si elle descend sous le seuil de +0.5, on estime qu'il n'y a pas de "match" et le tracking est abandonné.
Le rectangle externe correspond à la zone de recherche. (ce sera mieux expliqué dans le guide d'utilisation qui est en cours de rédaction smile)

1,642

(4 replies, posted in Bug reports)

Ok, thanks. That corresponds to the bug I was mentionning. The French locale was loaded by default instead of relying on system's.
This should be fixed in the next version.

1,643

(4 replies, posted in Bug reports)

Hi,

There is only one version, it has all the languages packed in.
The program first tries to detect if your system language is supported. If not, it should fall back to English.
However there might have been a bug that would make it fall back to French instead.

- Could you write here what the first menu is ("Fichier", "Tiedosto", something else ?)
- To change language, go to the second menu starting from the right, then first submenu.

Are you on an English Windows ?

1,644

(6 replies, posted in General)

Hi,
No, it is not currently possible.

It was necessary to improve the tracking performance and robustness before implementing multi-point tracking.
It is planned as a future improvement though.

1,645

(1 replies, posted in Bug reports)

Hi,
I hope someone can come and answer this better than me because I have not much experience with AVCHD, and I don't do a lot of conversions.

If you are not much concerned with storage size, you can use the highest quality option available in your conversion software.

Regarding blurriness/quality, bitrate is not as much relevant as the compression method used, because most of the time is spend decoding the image. This is especially true for large image size like full HD.
For a given codec and image size, you can use a higher bitrate. That should improve quality without degrading rendering speed. (You could use uncompressed files).

Regarding jerkiness, this will really depends on the machine performances and you'll have to make tests to find your threshold.
If you see the speed slider decreasing on its own, then the program can't keep the pace.
I would target for 25 or 30 fps, and then reduce size until it plays smoothly if you have jerkiness issues or automatic slow down.

Also, if possible try to use an intra-frame only codec (like DV or MJPEG) since that will make the backward navigation easier and decoding faster.

1,646

(1 replies, posted in Français)

Bonjour,

Pour que les menus soient activés, la video doit être en mode analyse: les images décompressées en mémoire.
Ce mode se déclenche tout seul quand la zone de travail est suffisamment courte, la valeur par défaut étant 12 secondes.

D'autres infos sont dispo dans l'aide aux chapitres Preferences et Explorer la technique.

Note: dans les nouvelles versions (0.8.0 et supérieures) la fonction miroir est disponible tout le temps.

1,647

(8 replies, posted in General)

Thanks guys big_smile

"Image sequence" is sort of taken by the image export tool when saving multiple images at regular intervals.

I really like "Overview" smile
I wonder if it would be clear enough alone, considering it's under the Motion menu. (So it would read Motion > Overview)

Multiple View sounds good too, but somehow slighltly less characteristic of the function ? (I fear users might mix it up with the two screen mode for video comparison)

1,648

(12 replies, posted in General)

Hi,
Thanks for the follow-up on the drag and drop and for the bug report.

I reproduce it too. It'll be known as bug m177.

1,649

(8 replies, posted in General)

First, the result of the function to rename in case you are not familiar with it :
It takes the working zone and creates a composite image out of it.

http://img13.imageshack.us/img13/8015/mosaic300.jpg

I guess "Mosaic" is a suboptimal name for this function…
- It doesn't really speak for itself…
- In recent years, the term has come to mean something else entirely: an image created from thousands of smaller images, each one acting as a pixel in the final composite.
- It recently came to my attention that a proprietary sport analysis package was using the term "Mosaic" for one of its function somewhat similar in concept than Kinovea's Mosaic. The company behind the package has an history of patenting and trademarking things, so I think it would be best to use a different name.

Naming a function is pretty hard, please put your ideas in the hat, comment on suggestions, etc.
- Should convey the idea of multiple images and/or of video summarization.
- Should be short enough (one / two words).
- Chronophotography is reserved for a future function.

Existing suggestions:
- Video Summary or Image Summary or Summary
- Synopsis
- Kinorama
- Composite

Thanks for your help big_smile

The discussion goes on over this thread: The tagging option... When ?