CVPR 2018: Learning to See in the Dark
Machine Learning algorithm that significantly improves quality of low light images

Follow by Email
Paper: Project page:


Milk Surface : This is amazing!

Victor Nikolov : I want to implement it in my glasses. :D

Alfredo Fernandes : I had an idea close to that after seeing the Nvidia denoiser for raytracing, shame I didn't had the knowledge to do such, but the results look really great!

Carl Ryu : Awesome work. No more YUV pipeline with denoising algo work. Good job

Darth Jahus : Science is so amazing!

Matthias Edling : Deep learning is the future and the future is now. Absolutly astonishing work, respect!

Graphic Solutions Xpress Network : Can this work with normal exposures, to correct shadow areas?

Benjamin Healy : Can this be applied to images that have already been compressed? It would be wonderful to be able to recover old shots of mine.

Fly-High : I want plugin for Lightroom!

Hartvik Line : Wow, that is freaking amazing!

Mary Shen : Your work is excellent.

Big John : The end of photography. AI will take over.

Jesper Sahner Pedersen : Wonder if and how this approach will work for videos with extra dimension(s) in terms of movement etc. Maybe future "looks bright" ;)

张志亮 : 好厉害 : *Wonderful technology !* If you want to, then people would fund you on crowd funding platforms for bringing this to phones. Let's get this implemented into the open camera on Android, and Gimp for image editing, so that we all can enjoy decent night time photography.

Dumb Jason : this is so ridiculously WTF. congratulations on this

Wan HL : 1.6s handheld is possible. Olympus em1ii has done it.

Prakash Pandy : How to pass the raw image from iphone camera to it?

Noe Miniel : this is amazing!

Joe Hrdina : This is amazing

Yanfei Wang : Wow, you guys just created the magic!

Rizki Adi Saputra : improve night vision

Simeon Radivoev : All we need now is temporal coherence

Nikhil Manohar : Stunning work!

Jason M : Is Chen this person's first name or last name? 🤔

Diamond Canvas : do you have any service? may pay, where able to use the standard designer?

sir pei : 凤梨酥看上去很好吃的样子

Lorenzo Betto : 2:13 how did it recognise the 'Digestive' writing on the right of the image?! The original image didn't give any clue about the presence of text.

rubixjohm : Amazing, great work!

Dmitry Vorobyev : Where I can download it for photoshop?

vishualee : I'm speechless :O

Nikita Nikiforov : it's a kind of magic! you're cool, guys

cfrek : This made me go "holy shit"

Daniel Holt : E N H A N C E

Graham : Great work, I'm amazed.

강동옥 : But there are no iso, shutter speed, aperture infos. I cannot trust the result before seeing the information

Tcim : next SOTA!

Kram1032 : Gotta say I'm surprised just how well Google's Pixel 2 camera does at the end at 15fps. - There are some artifacts but I'm almost certain those are from video compression which tends to disproportionately affect darker areas which is problematic if "darker" is the entire image.

Iluhandric : That's not good brother

R3ap3rPy : You are awesome!!!!!!

Закиров Марат : Could the same results be achieved on compressed data like JPEG -> JPEG? Or it strongly depends on RAW data of specific camera at CANNOT generalize at all?

Ashok Kumar Pant : Too good, amazing!!

magnusmoty : This is so very amazing! Will there be simple GUI software capable of doing this available to people?

Max Mescaline : Magic!

Sammy Hasan : Stunning!

Виталий Столяров : Amazing!

AJAY singh : great work

Andrew B : Very Impressive!

sdchew : Very impressive!