Hi,
I have a lot of collection that i made myself , like “The Best Of The 90’s” or "Dance Collection"
Off course that i can’t find any album that even comes close to my personal collections and i wondered how can i create manually an album (doesn’t have to be uploaded to Picard disc servers) locally on my machine so that Kodi would recognize it as one album and not 5345469475423!!@# different albums 
Thanks in advance …
Picard itself is not really suited for manual tagging, even if you can do it. You can of course edit the tags manually inside Picard for all tracks, so that they have the same album name etc. set. But to be honest there are tools better suited for this kind of manual work. You can of course use Picard to load the tags from MB and then fine tune them individually.
Regarding Kodi I don’t know which criteria it applies despite the MusicBrainz IDs to group an album together. I would probably try to remove all MusicBrainz IDs (most important the “MusicBrainz Release Id” and “MusicBrainz Release Group Id” from the files, if any, and make sure all files share the same album, albumartist and albumartistsort tags and are stored in the same folder.
Maybe somebody knows better, but it might be faster to ask in the Kodi forums 
EDIT: Updated to clarify which tags to remove and edit
Thanks for the quick reply, when i open a folder in Picard with a lot of files and than i try to search for suitable album that song is getting Acoustic ID and songs that were not recognized are missing that entry completely …
No quite sure if this is a question, but if you use the “Scan” function Picard uses acoustic fingerprinting to identify the songs using a system called AcoustId. If AcoustId cannot identify a song there is of course also no AcoustId that could be stored in the tags. But back to the topic of Kodi I am very sure that the AcoustId is absolutely irrelevant for Kodi and how it identifies files belonging to the same album 
And just to be complete: Instead of “Scan” you can also use “Lookup”, which instead of using AcoustId does a lookup based on existing tags.
I have also edited my above post to clarify which tags I meant you should remove and edit. But as I said before this is just a guess.
Regarding Kodi’s implementation of MusicBrainz. I have been following the progress of that (mostly lurking) and in recent times the development has moved on some and it is starting to be very usable.
I am not sure it can be done for your use case though.
There are many challenges required to get MusicBrainz support in a player and they are the first that know that have started to tackle it.
The best way for their implementation is an “all or nothing” approach. As Picard does support your approach to tagging - it requires matched releases - it is difficult for them to a coherent implementation.
You can try removing all the MBID’s and turning off all the scrapping functions but I am not sure how that will function. As they do also look at other non MBID tag feilds that are used (exclusively?) by picard, artists for example.
Cases like yours as one of the many challenges they are facing. People tagging with different versions of Picard is another (some tags generated are in some versions and not others) and some having some files with musicbrainz tags and not others is another (make everyone update to picard tags or not at all), etc as everyone has different music collections.
Adding full MusicBrainz and catering for everyone’s different collections and having a decent experience is a challenge. But one that they are embracing as a way forward and committed to using Musicbrainz I am glad to hear.
@guandms, if you “build” your own albums then you have little benefit from Picard or MusicBrainz. They are designed to match original released albums that you have purchased and ripped, to have them tagged correctly.
When you build your own albums for playback in Kodi, you should use a program like MediaMonkey, Mp3Tag og Foobar2000 to group songs into “albums” and tag them manually. Here’s what I do:
Album Name = Give the album a name that you like
Album Artist = Various Artists
Track # = 1…n (whatever order your choose. Don’t put more than 100 tracks in one album)
Title = Name of the song
Artist = Name of the artist playing (If more than one, separate them by semicolon or forward slash)
Then add an albumart of your liking, typically 500x500 jpg.
Tag the album as a compilation
Organise your albums in folders like this:
/Music/Various Artists/Album Name/track# - Artist name - Song Name.mp3 (or flc if you use flac)
Do not use APE tag format. Use ID3 v2.3 (Avoid 2.4 as Windows do not understand it)
Then tell Kodi to use embedded data when scanning your library and everything will show up nicely organised.
There is no point in feeding such “home made” albums to Picard cause it cannot provide much on an album level anyway. If you feed songs one-by-one to Picard, it will always try to match each song to an album from where it was ripped. Unless you want that information, it will only be confusing. If you do want it, it will be a tedious job if you have many songs.
Good luck