Honor AI camera discussion thread

SaiyanGoku

kamehameha!!
Answer
q1) Mosquitoes
Q2) Next feature must be include "the "score percentage" and " skill percentage" for taken image so that we can improve the image quality and also compare our score with rest of the world" and now everyone can be a professional photographer.
Q1: You want to identify mosquitoes on a phone camera?
Q2: "Professional" photographers don't depend on an "AI" camera to take pictures. Wannabe "professionals" can just use Modded gcam on auto HDR+ mode and if needed use snapseed to edit images.
 

Minion

Conversation Architect
If its AI then it should recognise everything.
Like if i focus my camera on a tree then it should tell me what tree it is.
Nope, if it recognizes tree or plant it will apply predefined settings on photos to make it look better

In simple words honor 10 will automatically post-processes images after recognizing what type subject
 

billubakra

Conversation Architect
My friend got the P20 lite, very good and accurate camera. It needs a mode like scanner or something so that we can take good quality scanned photos of stuff.
 

Charchit Sharma

Journeyman
AI is improving day by day, the features of Honor 10 has can be improved by updating the ML model architecture using deep learning models etc, Camera features can be updated as well, as Google is doing with their pixel, if you can improve the camera performance in single lens using AI and ML as they did, you can incorporate the feature for the dual camera's as well increasing the whole new performance level.
Except that, I think Honor 10 has every effect that is needed, and yes that feature can help using audio beep to indicate the photo is out of frame or not good.
 

Prasad Kharkande

Right off the assembly line
Participate in our Honor discussions, post your opinions, get likes on your posts and stand to win an Honor 10 worth ₹33,000 & up to ₹70,000 PayTM cash!

*www.hihonor.com/global/wp-content/uploads/2018/04/honor10005pc.jpg

During the launch of the Honor 10 on May 15, the company made a big deal about the way AI works its magic when taking pictures, enhancing the images taken by the camera setup even further.

In fact, the Honor 10 is more advanced than the AI used in the View 10 in 2 ways - Advanced Scene Recognition & Semantic Image Segmentation.

First off, is the more advanced scene recognition system that helps the phone’s AI recognise what exactly the camera looks at before it takes the shot. This lets the AI automatically apply the right settings to the picture before its taken.

So, the Honor View 10 could recognise 13 categories, which is pretty impressive. However, the Honor 10 can recognise 22. Some of the scenes include:
  • Fireworks
  • Waterfall
  • Cars
  • Cloud
  • Blue Sky
  • Nightsky
  • Flowers
What new scenes do you want to be added in the next Honor camera?

What extra features do you want Honor to incorporate in their next camera? Maybe, an audible warning when the photo doesn't look good? Or a refusal to take "duckfaces"?


Post your comments to the 2 questions in this discussion thread.

Honor is keeping a close eye on this thread and may even implement some of the suggestions mentioned over here.

Like we said earlier, post your comments and stand to win an Honor 10 worth ₹30K and up to ₹70,000 PayTM money. For more info, check here in a new tab for more details.

Participate in our Honor discussions, post your opinions, get likes on your posts and stand to win an Honor 10 worth ₹33,000 & up to ₹70,000 PayTM cash!

*www.hihonor.com/global/wp-content/uploads/2018/04/honor10005pc.jpg

During the launch of the Honor 10 on May 15, the company made a big deal about the way AI works its magic when taking pictures, enhancing the images taken by the camera setup even further.

In fact, the Honor 10 is more advanced than the AI used in the View 10 in 2 ways - Advanced Scene Recognition & Semantic Image Segmentation.

First off, is the more advanced scene recognition system that helps the phone’s AI recognise what exactly the camera looks at before it takes the shot. This lets the AI automatically apply the right settings to the picture before its taken.

So, the Honor View 10 could recognise 13 categories, which is pretty impressive. However, the Honor 10 can recognise 22. Some of the scenes include:
  • Fireworks
  • Waterfall
  • Cars
  • Cloud
  • Blue Sky
  • Nightsky
  • Flowers
What new scenes do you want to be added in the next Honor camera?

What extra features do you want Honor to incorporate in their next camera? Maybe, an audible warning when the photo doesn't look good? Or a refusal to take "duckfaces"?


Post your comments to the 2 questions in this discussion thread.

Honor is keeping a close eye on this thread and may even implement some of the suggestions mentioned over here.

Like we said earlier, post your comments and stand to win an Honor 10 worth ₹30K and up to ₹70,000 PayTM money. For more info, check here in a new tab for more details.
Now a days it’s a trend of mobiles have dual camera in which one of the camera captures the image whereas the other sense the depth on the image...

So I would suggest to have triple camera setup with AI imbedded into it. In triple camera setup we have first camera with low aperture second camera for depth sensing and third camera with high aperture.

So in this we can implement AI to sense the light around us ..So according it will choose the camera to capture the object. For example first camera with low aperture can be used for night photography with depth sense camera (2nd camera), third camera during day time with depth sense camera....

In this setup we can include machine learning where in a mobile camera tends to learn things according to user's usage...

AI can help to improve the image quality by having features like night mode, portrait mode, sports mode (to capture the moving objects)... (For those people who tend to auto settings of camera).

Manual mode should provide features like adjustment of shutter speeds, exposure, focus, ISO.... (Now a day some phones don’t provide the option of shutter speed adjustment.)

This was about the rear camera...

Now coming to front camera. It should include AI for face reconigation, and beautification modes which most of the phones have...

Last but not the least the phone should have a better battery performance with better camera performance...
 

ankushv

Journeyman
OT
How many of us actually use cameras that often .
Except functions or birthdays etc.
A better feature to have is a better low light camera f1.8 or lesser with optical image stabilization .
A better battery .
A dual 4g capable chipset .
Easy to use form factor .
Good speakers .
And the feature in the phone we use most often , a direct sunlight legible power effecient display .

Sent from my MI MAX 2 using Tapatalk
 

Prasad Kharkande

Right off the assembly line
No. Just no. This is a bad idea and ruins the image.
I agree with you...but most of the times some people tend to use beautification modes...So it may be not required feature for many of them...but keeping in mind the required feature by minority user i thing that company should also consider their requirements....
 

SaiyanGoku

kamehameha!!
I agree with you...but most of the times some people tend to use beautification modes...So it may be not required feature for many of them...but keeping in mind the required feature by minority user i thing that company should also consider their requirements....
If companies kept the requirements of "minority" users in mind, we won't have crap bloated crippled stock UIs or wafer thin phones with 2 hour battery life.

Sent from my ONEPLUS A3003 using Tapatalk
 

Prasad Kharkande

Right off the assembly line
If companies kept the requirements of "minority" users in mind, we won't have crap bloated crippled stock UIs or wafer thin phones with 2 hour battery life.

Sent from my ONEPLUS A3003 using Tapatalk
I'm saying about the camera and "not things like UI, Battery or some apps"
 

billubakra

Conversation Architect
:rofl: :rofl: :rofl: :rofl: :rofl:
Now that's the real world usage of AI. Hoping OEMs would market this feature instead of useless "beautify" mode and what not.
As per the NSA leaks CIA, Mossad and other spy agencies are using this stuff. Very unlikely that it will come to phobes.
 

Sajal Halder

Right off the assembly line
Scenes required:

* Fog scenario

* Dark mode, its different from Night mode. Here no light is present. Flash should be on all the time. then focusing on the desired subject should work properly.

*
High humidity mode, sometimes areas specially in beach or during rainfall the glass of lens becomes whitish from inside. Need to take proper action to make sure that doesn't happen or image shout be optimized specially for that type scenarios.

* Group mode. The dual picture we get from dual camera have limitations like the focus part is small. During focusing other body parts gets blurred which is suitable only for close photos or portraits. But in this mode phone should recognize all the subjects then takes multiple photos taking different body parts in focus and then merge them into one picture so that all parts of subjects. Or we can manually add the places of focus for head arms legs like the way we d it with eyes and lips in several beauty apps. This way we can get a step closer to the pictures taken with the original DSLR camera.

* Single subject DSLR mode. This is similar to the above mode but the point of focuses are of single subject. Again closer to DSLR camera.

* During group picture ( no groupie), front camera should also take pictures and there should be an option to position his face in the group. like between two people.

For Video capture

* 4K or full HD slo mo video capture at high FPS. Also, user must be allowed to mark the portion of the video which he want to see in slo mo and he must be allowed to convert it to other video format. like eg. mp4.

* Sync mode. This is a very cool and unique mode I would like to suggest. In this mode, the FPS of capturing the video should be synced with the subject in such a way that it would look still. For eg, wheels of vehicles, wings of bird. AI should understand the part of subject (manual is good to). After capturing the video the wheels of car would look still but the car is still moving. or the bird is flying but the wings are always open and stationary. ( I know their speed varies so after capturing, processing is required to eliminate blurred and different frames from the final video). Must be allowed to convert it to other video format. like eg. mp4.

*
Still mode. another cool mode: Here the whole frame should be still only the moving parts of the subject should be moving. Like the the car, road, trees ..all are stationary objects. But the wheels of the car is moving. A candid campfire scene where the everyone is stationary like except the fire. This can be done if AI or the user can select the portion of the video which should be moving and rest the parts can be selected from any suitable frame of the captured video. Must be allowed to convert it .gif format.

* AI should recognize all the above scenario. Or at least give prompt to apply suitable scenario/mode.
* In the main display where we press the shutter button AI should show live preview in a small window showing the output might come after applying the AI suggested mode. Without disturbing the current scene I am capturing.
* AI should suggest to tag the people captured in the video or picture in social media by recognizing their face.
* AI should understand and suggest not to take identical pictures.
*AI must allow us to use multiple modes in single picture.
* AI should show us the best possible outcomes of a picture if edited.

 

SaiyanGoku

kamehameha!!
* AI should recognize all the above scenario. Or at least give prompt to apply suitable scenario/mode.
* In the main display where we press the shutter button AI should show live preview in a small window showing the output might come after applying the AI suggested mode. Without disturbing the current scene I am capturing.
* AI should suggest to tag the people captured in the video or picture in social media by recognizing their face.
* AI should understand and suggest not to take identical pictures.[/SIZE]
*AI must allow us to use multiple modes in single picture.
* AI should show us the best possible outcomes of a picture if edited.
You need Jarvis and/or Iron Man's armor. A phone's AI isn't designed for these many complicated tasks at once.
 

Sajal Halder

Right off the assembly line
You need Jarvis and/or Iron Man's armor. A phone's AI isn't designed for these many complicated tasks at once.

I was suppose to give ideas about camera AI. Having Jarvis is like having an assistant who works not only camera. (By the way, Marvel got idea of Jarvis from Alfred of Batman, just modified from human to AI)
 
Top Bottom