Inpainting ati Outpainting adanwo

Awọn idahun ati awọn alaye han bi o ṣe n mu ibeere naa.

1. Kí ni inpainting ṣe?
   • Ṣe atunto agbegbe ti o boju-boju ninu aworan kan
   • Fa aworan kan kọja awọn aala rẹ
   • Iyipada ọrọ si ọrọ
   • Compresses image awọn faili
2. Kini iṣẹ ita gbangba ṣe?
   • Yọ awọ kuro lati awọn fọto
   • Ṣe ipilẹṣẹ akoonu tuntun ju awọn egbegbe atilẹba lọ
   • Ṣe awari awọn oju
   • Pọ awọn piksẹli blurry
3. How does RePaint use the known region during diffusion-based inpainting?
   • It permanently freezes every unmasked pixel with no resampling
   • It removes the mask before denoising
   • It repeatedly resamples and conditions generation on known pixels during reverse diffusion
   • It trains a new model from scratch for each image
4. What architecture does LaMa introduce for large-mask image inpainting?
   • A language-only transformer
   • A nearest-neighbor search over captions
   • A fixed-color fill with no learned model
   • Fast Fourier Convolutions with a receptive field that supports image-wide context
5. Kini itọsọna akoonu ti o han ni agbegbe ti o boju-boju?
   • Titọ ọrọ ati awọn piksẹli agbegbe
   • Orukọ faili nikan
   • Aago ẹrọ
   • Ipele funmorawon aworan
6. Kini idi ti ita gbangba ṣe wulo fun atunṣe?
   • O paarẹ koko-ọrọ naa
   • O le yi ipin abala aworan pada nipa fifi iwoye kun
   • O encrypts aworan
   • O dinku ipinnu
7. Besides Fourier Convolutions, what two components does the LaMa paper identify in its large-mask inpainting method?
   • A speech decoder and image captioner
   • A fixed image palette and nearest-neighbor lookup
   • A high receptive-field perceptual loss and large training masks
   • A video tracker and optical-flow classifier
8. What does a mask specify in an inpainting task?
   • The image’s camera model
   • The order of training examples
   • The output file’s copyright status
   • Which image region is to be modified or synthesized