ImageX - AI Photo Enhancer 4+

Name: ImageX - AI Photo Enhancer
Rating: 3.5 (6 reviews)
Author: 嘉豪 吕

嘉豪吕

- 3.5 • 6개의 평가

- 무료
- 앱 내 구입 제공

설명

- ImageX built-in a variety of latest image enhancement AI algorithms.
- AI Super Resolution, enlarges the image, improve the image quality, noise reduction.
- AI Frame Interpolation, improve the smoothness of animated image.
- Dozens of pre-trained models, suitable for different styles.
- Local offline processing, no upload required, ensuring privacy.
- Support GPU hardware acceleration, faster processing speed.
- Rich format, support JPG PNG APNG GIF WebP HEIC format.
- No ads.
- Supports iOS and Mac (sold separately).
(For batch processing and faster speeds, we recommend the Mac version)

Privacy Protection:
All images are processed offline locally on the device, no uploading, no privacy risk.

Non-Pro version restrictions:
- iOS
1. Any side of the output that is higher than 1280 pixels or the output that is an animated image does not support saving and sharing.
2. Batch saving and sharing are not supported.
- Mac
1. Any side of the output that is higher than 1280 pixels or the output that is an animated image will be ignored.

Subscription:
1. Upgrade to Pro with Subscription or Lifetime Purchase to unlock full access.
2. If subscribing to the Pro version (monthly subscription or annual subscription; lifetime purchases are not included), the subscription will automatically renew unless you cancel the subscription 24 hours before the end of the next recurring date.
3. Pro for iOS and Mac are sold separately; iOS in-app purchases are only applicable to iOS App use, and Mac in-app purchases are only applicable to Mac App use.
4. For batch processing and faster speeds, please choose the Mac version.

Privacy Policy: https://www.tinyperfectapps.com/res/imagex/policy/imagex_privacy_policy.html
Terms of Use: https://www.tinyperfectapps.com/res/imagex/policy/imagex_terms_and_conditions.html

Reference projects:
Waifu2x
- nagadomi/waifu2x: https://github.com/nagadomi/waifu2x (MIT)
- lltcggie/waifu2x-caffe: https://github.com/lltcggie/waifu2x-caffe (MIT)
- nihui/waifu2x-ncnn-vulkan: https://github.com/nihui/waifu2x-ncnn-vulkan (MIT)
Real-CUGAN
- bilibili/ailab: https://github.com/bilibili/ailab/tree/main/Real-CUGAN (MIT)
- nihui/realcugan-ncnn-vulkan: https://github.com/nihui/realcugan-ncnn-vulkan (MIT)
Real-ESRGAN
- xinntao/Real-ESRGAN: https://github.com/xinntao/Real-ESRGAN (BSD3)
- xinntao/Real-ESRGAN-ncnn-vulkan: https://github.com/xinntao/Real-ESRGAN-ncnn-vulkan (MIT)
SRMD
- cszn/SRMD: https://github.com/cszn/SRMD (MIT)
- cszn/KAIR: https://github.com/cszn/KAIR (MIT)
- nihui/srmd-ncnn-vulkan: https://github.com/nihui/srmd-ncnn-vulkan (MIT)
RealSR
- jixiaozhong/RealSR: https://github.com/jixiaozhong/RealSR (Apache2)
- nihui/realsr-ncnn-vulkan: https://github.com/nihui/realsr-ncnn-vulkan (MIT)
RIFE
- megvii-research/ECCV2022-RIFE: https://github.com/megvii-research/ECCV2022-RIFE (MIT)
- nihui/rife-ncnn-vulkan: https://github.com/nihui/rife-ncnn-vulkan (MIT)
IFRNet
- ltkong218/IFRNet: https://github.com/ltkong218/IFRNet (MIT)
- nihui/ifrnet-ncnn-vulkan: https://github.com/nihui/ifrnet-ncnn-vulkan (MIT)

2023년 8월 26일

버전 1.2.0

1. Add shortcut for import images.
2. Configuration can be stored.
3. The output directory is also stored.
4. Bug fixes.

3.5/5

6개의 평가

Great model selection, just needs video support

There isn’t much of a selection yet on mobile devices for AI upscaling or interpolation models, just Waifu2x and a couple other random little apps. I think the capabilities and speed of these models will make them commonplace one day, so I’m excited to see a developer working on implementing Rife and other models into their app.
I know it’s not an easy ask, but Rife 4.6 for video files would be AMAZING. It’s the best interpolation model I’ve used, and FlowFrames (windows app) gives almost real-time speed with the VapourSynth implementation. Exciting stuff! Hope to see this app continue to grow and evolve.

Update

I originally gave this 5 stars but there needs to be a downscale option so that you can limit the upscale

We did attempt to incorporate a pre-downscale feature during the development stage. However, despite subsequent upscaling, the image quality deteriorated significantly.