Fooocus 2.2.0 Update – Prompt Array | Metadata | & More! (Stable Diffusion)
Summary
TLDRこの動画は、Focus更新プログラムの重要な変更点について説明しています。主な変更点は、出力ファイル形式の選択肢が増えたこと、Loraのオン/オフ切り替えが簡単になったこと、Loraの重み付け範囲が拡大されたこと、中間結果の非表示オプションが追加されたこと、固定シードでの連続生成がしやすくなったことなどです。また、プロンプトにおける配列入力のサポート、メタデータの画像への埋め込み機能、モデルとLoraのパスを複数指定できるようになった点も紹介されています。これらの改善により、ユーザーの利便性が大幅に向上したことがわかります。
Takeaways
- 📌 このビデオでは、Focus更新プログラムの重要な改善点について説明しています。
- 🖼️ 新しい出力フォーマット(PNG、JPG、WebP)がサポートされるようになりました。
- ✅ LORAsのチェックボックスが追加され、簡単にオン/オフできるようになりました。
- ➕ デフォルトのMax LORAsの重みを+/-2を超えて設定できるようになりました。
- ⭐ プリセットを使って設定を保存・変更することが推奨されています。
- 🚫 「disable intermediate results」を有効にすると、複数画像の生成中に完了した画像が表示されなくなります。
- 📝 「disable seed increment」を有効にすると、配列プロンプトを使って同一シードで複数の画像をレンダリングできます。
- 💾 「save metadata to images」を有効にすると、メタデータが画像に保存され、後で設定を復元できます。
- 📂 モデルとLORAsの追加パスを設定ファイルに追加できるようになりました。
- 📝 設定ファイルの変更は、プリセットを使って行うことが推奨されています。
Q & A
この動画では、どのようなアップデートについて説明されていますか?
-この動画では、Focus (AI画像生成ツール) の新しいアップデートについて説明されています。特に、要求の多かった重要な改善点に焦点を当てています。
新しいアウトプットフォーマットはどのようなものですか?
-以前はPNG形式のみでしたが、新しくJPGとWebPの出力フォーマットがサポートされました。これにより、ファイルサイズを小さくすることができます。
LORAsを有効/無効にする新機能について説明してください。
-LORAsの横にチェックボックスが追加され、LORAsを簡単に有効または無効にできるようになりました。これにより、LORAsの重みを0にする必要がなくなり、より手軽に切り替えられます。
最大LORAs重みの設定範囲が拡張されたのはどのような理由からですか?
-一部のLORAsが+3や-3などの高い重みを必要とするため、従来の範囲では不十分でした。この変更により、そうしたLORAsを完全に活用できるようになりました。
プリセットの作成方法について説明されていますか?
-はい、プリセットの作成方法が簡単に説明されています。config ファイルを直接編集するのではなく、プリセットを作成して設定を変更することが推奨されています。
新しい「disable intermediate results」機能の目的は何ですか?
-この機能を有効にすると、複数の画像を生成する際、現在生成中の画像のみが表示され、完了した画像はプロセス終了までサイドに表示されなくなります。
「disable seed increment」機能と新しい配列プロンプトの関係を説明してください。
-この機能を有効にすると、同じシードで配列プロンプトを使って複数の画像を生成できます。これにより、色やテキストの変化のみを比較することができます。
メタデータの保存機能についてどのように説明されていますか?
-生成した画像にメタデータ (使用した設定など) を埋め込むことができるようになりました。この機能は長らく要求されていましたが、プライバシーの観点から搭載が遅れていました。
複数の場所にあるモデルやLORAsをどのように扱えるようになりましたか?
-config ファイルにパスを追加することで、複数の場所にあるモデルやLORAsを登録できるようになりました。この変更により、より柔軟な運用が可能になります。
この動画全体を通して、どのような主なメリットが強調されていますか?
-この動画では、長年要望されていた機能が実装されたことで、ユーザーの利便性が大幅に向上したことが強調されています。ツールの柔軟性と効率性が高まったと言えるでしょう。
Outlines
👨💻 Focus 更新で追加された主な新機能
この動画では、Focus の最新アップデートで追加された主な新機能について説明しています。まず、画像出力形式として PNG に加えて JPG や WebP が選択できるようになりました。次に、Loras のチェックボックスが追加され、Loras の有効/無効を簡単に切り替えられるようになりました。また、Loras のデフォルト最大重み付けが +/- 2 を超えて設定可能になり、高い値の Loras の活用が可能になりました。さらに、複数の Loras を使用できる上限値を増やすこともできます。
🔧 その他の新機能と設定
この動画の後半では、その他の新機能と設定変更方法について説明しています。新しい「disable intermediate results」と「disable seed increment」の設定が追加され、中間結果の表示方法やシード値のインクリメントをコントロールできるようになりました。また、テキストプロンプト内で配列構文を使って、一度に複数のバリエーションを生成することが可能になりました。さらに、画像メタデータの保存機能が追加され、画像生成時の設定を画像に埋め込むことができるようになりました。最後に、モデルとLoras の複数のパスを設定ファイルで指定する方法を紹介しています。
Mindmap
Keywords
💡出力フォーマット
💡Aura
💡最大/最小重み
💡プリセット
💡デバッグ機能
💡配列プロンプト
💡シード固定
💡メタデータ保存
💡パス指定
💡更新の必要性
Highlights
The Focus update introduced important improvements that have been requested for a while.
New output formats like JPG and WebP were added in addition to PNG.
Checkboxes were added to easily activate or deactivate LORA without changing weights.
Max weights for LORAs can now go beyond +/- 2.
Presets can be used to modify config settings like max LORA weights and number of LORAs.
New option to disable intermediate results and only show the current generation image.
New option to disable seed increment, allowing generation of multiple images on the same seed.
Support for arrays in prompts, enabling generation of multiple unique images with variations.
Arrays in prompts can be used with disabled seed increment to compare variations on the same seed.
New option to save metadata to images, allowing easy reloading of settings from the image.
Metadata includes more information than the history log.
Multiple paths for models and LORAs can now be added to the config file.
Instructions on modifying the config file for presets and adding multiple paths.
Explanation of how arrays in prompts work and examples of their usage.
Recommendation to create and modify presets instead of directly editing the config file.
Transcripts
in this video I want to talk about the
new Focus update I normally don't cover
the updates but this made some important
improvements that have been requested
for a while and I felt it was worth
going over there are several small
changes that won't concern or even be
noticed by most but we do have a few
nice quality of life improvements that
we have been waiting for if you want to
look at the official release notes I
will leave that link in the description
and you can check those out on the
GitHub page now let's open up our focus
and go into the advanced tab first major
change is the new output format before
we only had PNG option but now you can
also select jpg or webp while we could
have a whole other conversation on which
one of these to use if you don't know
yourself then PNG will be fine unless
you are really concerned about file
sizes as it is a lossless format meaning
there is no compression jpg and webp are
compressed and will be smaller but lose
some detail and if you are downloading 6
gig models without concern then using
PNG will be fine moving on to the model
tab we will see checkboxes next to the
ls has been added now we can easily
activate or deactivate Aur without
having to remove it or turn the weight
to zero if we wanted to shut it off a
very helpful change when you want to
quickly deactivate and then reactivate
Aura without losing your weight settings
this could get tedious when dealing with
multiple luras when trying to test
things out now another change here is
the ability to have the default Max
weights for Laura's go Beyond positive
and negative2 most of the time you
wouldn't want to go beyond these numbers
but there are luras created out there
that do go higher a big example is the
detail tweaker Xcel it is designed to go
as far as plus or minus three so up
until now we couldn't utilize it
entirely but we will have to change
these settings ourselves manually in the
past I would change the config text file
directly but I would recommend creating
and making changes in presets only now
mostly because the config file
occasionally gets overwritten plus it is
good to leave the default settings alone
if you don't know about presets I will
add a video in the description showing
how to create those so I have my old and
new custom preset here which is just a
copy of the default settings with my
changes you can see some new lines that
I have added for the new min max Laur
settings we have these two lines here
just to make it easier you can copy
these lines from the config modification
tutorial text file in your focus folder
so you don't have to try to type them
out yourself once added you can put
whatever numbers you want I will use
four for now but if I encounter aura
that needs to go higher I can always
come back and change this there is also
this additional line here this allows us
to use more luras than the standard five
I have yet to use more than five but you
can easily add this line in too and
change the number to whatever you like
this line can also be found in the
modification text file now back in Focus
I want to go to the advanced Tab and
then into
debug down at the bottom we will notice
three new boxes disable intermediate
results will make it so when generating
multiple images you will only see the
image currently being generated and will
no longer see completed images on the
side until the entire process is
done this isn't something useful to me
but I suppose if you would rather see
the current Generation image in the full
window than this is for
you then we have disable seed increment
what this does is when generating more
than one image even on a fixed seed the
next image will go up a number in the
seed otherwise all the images will be
exactly the same so how can we use this
well that is where another new feature
comes in and that is the new support for
arrays how this works is similar to
creating mini wild cards inside the text
prompt so let's say we wanted four
pictures of a woman with four different
color shirts in one prompt so how would
we write that using the new array
support we would type a woman wearing a
then for the colors we would would
create two brackets then write our four
different colors like this then end it
with two brackets and then type t-shirt
this can be two colors or as many colors
as you want you just need to enable
enough image generations to get all
iterations if you have four colors we
want four Images the other difference
between this and wild cards is this will
produce results in order of the input
with wild cards it will choose words in
a random order so doing it this way you
will always get a red green blue black
t-shirt in that order with this prompt
this can also be used multiple times in
the prompt such as this
prompt and make sure your image number
is high enough for the
prompt this will create 12 Unique Images
it might be confusing at first how these
generate but it will go blond Man red
brunette Man red then blond woman red
brunette woman red then blonde man gream
and so
on so you can create multiple unique
items this way very quickly without
having to create a wild card every
time and this isn't limited to single
words either you can use entire phrases
like this you just need to remember if
you have multiple combinations of arrays
that you will need to bump up the image
number uh to compensate if there are 10
possible images but I leave the image
number at two it will only ever show the
first two
combinations now getting back to the
disable seat increment setting with this
enabled it will allow you to use the new
array prompts and create all these
images on the same seat which is
beneficial when trying to compare the
differences without the image changing
too much now with all of these images on
the same seed we get very similar images
with just a shirt change it's not always
this similar but the idea is you are
seeing the same image on the same seed
with only the color
changed without this feature each of
these images are going to be the next
seed and you can see the differences are
more pronounced since they are different
seeds this setting really only works
with the array prompts right now as far
as I can tell even wild cards won't
change on the same seed you will just
get the same image over and over so keep
that unchecked
otherwise the last setting is the save
metadata to images this is useful if you
have been wanting your images to save
the info onto the PNG in order to easily
reload the settings you have used from
the image this was a feature not
included for a long time as the
developer didn't want it included to
protect data privacy of others but since
it was such an in demand feature it was
finally included as an option what it
does is very similar to the history log
but now it just attaches the information
directly onto the image itself so if you
have this enabled and create an image
you can now go to your input image menu
and you will see another new tab here
called metadata and you can load and or
drag and drop an image here as long as
it has metadata enabled you should see
this info with this you can apply the
metadata and it will load all the
settings used to create this
image this is just like using the
history log and using the copy to
clipboard pasting into the text prompt
then hitting load
parameters the metadata does hold more
information than the history log I have
noticed if you set custom steps for
something like lightning the history log
doesn't load that parameter but
otherwise is pretty similar so let's go
back to our preset documents and you can
see at the very very end we have three
new lines this is only if you want to
have metadata on by default same as
before you can copy these lines from the
modification text file have this set to
true and make sure not to capitalize it
metadata scheme is just the format which
it is saved keep it at focus in the last
line you can add a name if you want to
tag your Creations but keep it blank if
you are concerned about privacy you
actually can drag in an image created in
automatic 1111 or Forge to see the
metadata here but it won't be able to be
orted the last thing to mention is for
those who have models and luras in
multiple places you can now add all
those paths to the config
file to start this I will add a bracket
here then a
space add a parenthesis and paste in the
extra location for my
models add another parentheses a bracket
to close it and a comma also the back
slashes have to be double so you will
have to add those in and you can leave
it like this and it should work but if
you want it to look more organized you
can space some things out just hit enter
here and tab then go over here and hit
enter and
tab and then between the last
parentheses and bracket hit enter and
tab it's not necessary but makes it a
bit easier to read you can do the same
for the luras if you have extra in a
different location
too once you save that you can start or
refresh your focus and in the model tab
the new models from the secondary
Position will actually start at the
bottom of the list in case you can't
find them and I think that wraps this
one up guys hope this was helpful and if
you have any questions you know what to
do see you in the next one
Voir Plus de Vidéos Connexes
【公式発表!?】2024年インスタ最新情報
【要約】ジェームズ・クリアー式 複利で伸びる1つの習慣【ジェームズ・クリアー】
Logseq 0.10.7 Changelog - Disable accent colors
WordPress 5.3 Overview & NEW Features of the Block Editor
How To Use Templates – Bootcamp – Step-By-Step
⚠️警告⚠️知らないと収益停止⁉️AI動画で稼ぎたい人必見❗️YouTubeにAI規制が入りました❗️対策方法を超初心者向け徹底解説❗️【YouTuber】【AI副業】【AI動画生成】
5.0 / 5 (0 votes)