Can compile a Windows exe or DLL file and directly use other tools to obtain the results

[NOBB2333]

god ermig1979,

i wait so long , almost 3years, i wait the command line direct get result ,
i just want ask can add a release , my vs 2022 open Remind Me this Error failed to load the project file. Could not find path , Part of AmxBf16.vcxproj

i also follow AntiDupl to install Install vcpkg :

---

god ermig1979,

Я так долго ждал, почти три года, ждал, пока командная строка получит результат,
Я просто хотел спросить, можно ли добавить версию, которую я открываю vs 2022, чтобы напомнить мне, что этот « Error не смог загрузить проектный файл». Не найден путь, часть AmxBf16.vcxproj
Я также установил «install vcpkg» в AntiDupl:

[ermig1979]

Hi! I don't fully understand issue. Is there any MSVS-2022 project build error? I just have cloned AntiDuplX and built it successfully with using of MSVS-2022. Or do you mean something another issue?

[NOBB2333]

reply

Perhaps it's a configuration issue with my computer.

I just found out that AntiDuplX/3rd/Simd | Cpl | libjpeg turbo is a link that will redirect to your other repository
Subsequently, I downloaded these repositories and completed the compilation,

One more thing, in Windows, I just find only single quotes can be used in PowerShell, maybe options problem, I still find it in the origin code(LOLOLOLOL)

compile over

# origin
./AntiDuplX -id=./pict_dir_1 -id=./pict_dir_2 -ct=0.05 -of=./dupl_list.txt

# can use
./AntiDuplX -id='./pict_dir_1' -id='./pict_dir_2' -ct=0.05 -of='./dupl_list.txt'

PS: compilation is so big, 1.72G

final question

Finally, is this algorithm different from AntiDupl? I set -- compareThreshold=0.95, but many of them cannot be detected. I use the hashType in the source code to calculate

One more thing, can export result.txt add Modify delimiters or configure different delimiters ，sometimes the path have the space , result use not good

Thanks , problem solved

[ermig1979]

The image compare algorithm used in AntiDuplX is close to used in AntiDupl. There are slightly differences mainly in using of multithreading.

It will be good if If you can give me small example of wrong searching to local reproducing.

The using of other separator in results output is a good idea.
Tab symbol is seems good candidate. Of course I can add special parameter for this purpose.

[NOBB2333]

I set all test image and result in a zip, link in there https://github.com/NOBB2333/nobb2333.github.com/releases
which have 7 pic,

• pict_dir_1 (4 pic)
• pict_dir_2 (3 pic)

pict_dir_2 is full copy from pict_dir_1 ,
in pict_dir_1 i have a little cut in first pic, from PIC1 1280X720 to PIC1 COPY 1231x720

and another interesting thing is made ct=1.00 will get 6 result, result in AntiDuplX_dupl_list -- ct=1.00.txt

./AntiDuplX -id='./pict_dir_1' -id='./pict_dir_2' -ct=1.00 -of='./dupl_list.txt'

[ermig1979]

Hi. I Can't find reference to archive with test images.

[NOBB2333]

sorry ,my apologies, it's draft ,it's first times i use this publish a releases,

let's try this, i test this in another computer, dowload is ok

https://github.com/NOBB2333/nobb2333.github.com/releases/tag/AntiduplX

[ermig1979]

I reproduced this behaviour on another set of images.
I clarified AntiDuplX algorithm. It slightly differ from used in AndiDupl:

1. Use by default 16x16 image hash size (original AntiDupl uses 32x32). It is reasonable to use the old size 32x32.
2. For image resizing AntiDuplX uses Area method instead of Bilinear which is more precise. It reasonable to use new option.
3. AntiDuplX is designed for auto deleting of simular images so it hides some results. In other words if we have 3 simular images a.jpg, b.jpg, c.jpg and by quality a.jpg <= b.jpg <= c.jpg then output results will be only (a.jpg and b.jpg) and (b.jpg and c.jpg). The result (a.jpg and c.jpg) is skipped.

[ermig1979]

I performed next changes: updated Cpl, Simd, libjpeg-turbo; changed default compareSize (16x16 -> 32x32); began use tab symbol as results separator.

[NOBB2333]

thanks, i know more about this item, seems i need to study C++ now